Defining "Rural" Areas: Impact on Health Care Policy and Research

July 1989

NTIS order #PB89-224646 FOREWORD

The problems of health care in rural areas have long occupied a special niche in policies designed to advance the Nation’s health. Programs for recruitment, training, and deployment of health care personnel, for constructing health-care facilities, and for financing health care often have included special provisions for rural areas. These programs have often also included attempts to mitigate the negative impacts on rural areas of policies primarily designed for and responsive to the needs of urban areas. However, some rural areas continue to have high numbers of hospital closures, ongoing problems in recruiting and retaining health personnel, and difficulty in providing medical technologies commonly available in urban areas. Mounting concerns related to rural residents’ access to health care prompted the Senate Rural Health Caucus to request that OTA conduct an assessment of these and related issues. This Staff Paper was prepared in connection with that assessment.

Rural definitions may greatly influence the costs and effects of health policies, because the size and composition of the U.S. rural population and its health care resources vary markedly depending on what definitions are used. There is no uniformity in how rural areas are defined for purposes of Federal program administration or distribution of funds. This paper examines dichotomous designations used to define rural and urban areas and discusses how they are applied in certain Federal programs. In addition, several topologies are described that are useful in showing the diversity that exists within rural areas. These topologies may be helpful in identifying unique health service needs of rural subpopulations.

A second OTA paper, Rural Emergency Medical Services, will also precede the publication of OTA’s full assessment on Rural Health Care. Defining “Rural” Areas: Impact on Health Care Policy and Research

Maria Hewitt

Health Program Office of Technology Assessment Congress of the United States Washington, D.C. 20510-8025

July 1989

This Staff Paper is part of OTA’s assessment of Rural Health Care

Carol Guntow prepared this paper for desk-top publishing.

The views expressed in this Staff Paper do not necessarily represent those of the Technology Assessment Board, the Technology Assessment Advisory Council, or their individual members.

For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, DC 20402 CONTENTS

Chapter Page 1. Summary ...... 1

2. Introduction ...... 3

3. Delineating “Rural” and “Urban” Areas ...... 5 U.S. Bureau of the Census ...... 5 The Office of Management and Budget: Metropolitan Statistical Areas ...... 7

4. Relationship Between Urban/Rural and Metropolitan/Nonmetropolitan Designations ...... 13

5. Understanding Diversity Within Rural Areas: Urban/Rural Topologies ...... 17 Topologies Used To Describe Nonmetropolitan Areas ...... 17 Urbanization/Adjacency to Metropolitan areas ...... 19 Adjacency to Metropolitan Areas/Largest Settlement Size ...... 20 Population Density: Incorporation of the Frontier Concept ...... 20 Urbanization/Population Density ...... 21 Distance From an MSA or Population Center ...... 21 Commuting-Employment Patterns ...... 22 Economic and Socio-Demographic Characteristics ...... 22 Conclusion ...... 24

6. Availability of Vital and Health Statistics for Nonmetropolitan Areas ...... 27

7. Using OMB and Census Designations To Implement Health Programs ...... 35 Medicare Reimbursement Using MSAs To Define Urban and Rural Areas...... 35 The Rural Health Clinics Act ...... 37 Providing Services in “Frontier” Areas ...... 38

8. Conclusions ...... 41

Appendix A. Summary of the Standards Followed in Establishing Metropolitan Statistical Areas ...... 43 B. The Census Bureau’s Urbanized Area Definition ...... 47 C. Census Geography ...... 48 D. Rural Health Care Advisory Panel ...... 53 E. Acknowledgments ...... 54

References ...... 56

Tables Table 1. Urban and Rural Population by Size of Place (1980) ...... 6 2. Ten States With the Largest Rural Population (1980) ...... 7 3. States With More Than One-Half of Their Population Residing in Rural Areas (1980) ...... 7

iii CONTENTS (cent’d)

Tables (cent’d) Table Page 4. Ten States With the Largest Nonmetropolitan Population (1986) ...... 10

5. States With More Than One-Half of Their Population Residing in Nonmetropolitan Areas (1986) ...... 10 6. Selected Federal Department/Agencies Using MSA Designations for the Administration of Programs or the Distribution of Funds ...... 11 7. Population Inside and Outside of MSAs by Urban and Rural Residence (1980)...... l5 8. Classification of Nonmetropolitan Counties by Urbanization and Proximity to Metropolitan Areas ...... 19 9. Nonmetropolitan County Population Distribution by Degree of Urbanization and Adjacency to an MSA (1980) ...... 20 10. U.S. Population by County’s Largest Settlement and Adjacency to an MSA (1980) ...... 20 11. Bluestone and Clifton County Classifications Based on Urbanity and Population Density ...... 21 12. Population-Proximity: A Measure of a County’s Relative Access to Adjacent Counties’ Populations ...... 22 13. County Typology Based on Employment, Commuting, and Population Characteristics ...... 23 14. Distribution of U.S. Counties by Typology Based on Employment, Commuting, and Population Characteristics (1986) ...... 24 15. Classification of Nonmetropolitan Counties by Economic and Socio-Demographic Characteristics ...... 25 16. Features of the Nine County-Based Topologies ...... 25 17. Proportion of the Population 65 and Older by Metropolitan/Nonmetropolitan and Urban/Rural Residence ...... 27 18. Proportion of Nonmetropolitan Population Age 65 and Older by Level of Urbanization and Adjacency to an MSA (1980) ...... 27 19. Nonmetropolitan Infant Mortality Rates by Urban Area and Race, U.S. Total and Alabama (1986) ...... 28 20. Characteristics of Different Categories of U.S. Nonmetropolitan Counties ...... 30

Figures Figure Page 1. Urbanized Areas ...... 5 2. Metropolitan Statistical Areas (June 30, 1986) ...... 9 3. The Relationship Between Metropolitan Statistical Areas (MSAs), Urbanized Areas, and Urban and Rural Areas ...... 13 4. Map of California Counties: San Bernardino County ...... 14 5. 1980 Population Distribution (United States) ...... 18 6. Areas With Cervical Cancer Mortality Rates Significantly Higher Than the U.S. Rate, and in the Highest 10% of all SEA Rates (White Females, 1970-1980) ...... 31 7. Death Rates Due to Unintentional Injury by County ...... 32 8. Death Rates Due to Motor Vehicle Crashes by County ...... 33 9. Frontier Counties: Population Density of 6 or Less ...... 39 iv 1. SUMMARY

It is difficult to quantify rural health by the same agency. For example, Congress problems and to make informed policy deci- directed the Health Care Financing Adminis- sions without a clear definition of what and tration to use Census’ nonurbanized area where “rural” areas are. Small population, designation to certify health facilities under sparse settlement, and remoteness are all fea- the Rural Health Clinics Act, but to use tures intuitively associated with “rural. ” OMB’s MSA/nonMSA designations to cate- These features exist on a continuum, how- gorized hospitals as urban or rural for pur- ever, while Federal policies usually rely on poses of hospital reimbursement under Medi- dichotomous definitions. care. In general, rural hospitals are reimbursed less than their urban counterparts. Urban and rural areas are often defined While persistent differences between metro- using the designations of either the Office of politan and nonmetropolitan hospital costs Management and Budget (OMB) or the have been observed, hospital location may be Bureau of the Census. Rural areas are the a correlate rather than a determinant of cost remaining areas that are not captured in ei- differences. Therefore, hospital-specific ther OMB’s “metropolitan statistical area” measures are being sought that might replace (MSA) designation or in Census’ urban or ur- the present MSA adjustments to the basic banized area definitions. Counties are the prospective payment formula. Topologies building blocks of OMB’s MSAs and are easy that categorize counties according to their de- to use, because county-based data are readily gree of urbanization or their employment and available. One or more counties form an commuting patterns could be used to refine MSA on the basis of population size and the definition of labor market areas, an im- density, plus the degree of area-wide eco- portant component of the Medicare formula. nomic integration as reflected in commuting patterns. The Census’ urban and urbanized There have been calls to develop a stan- area definitions rely on settlement size and dard rural typology that would capture the density without following county boundaries, elements of rural diversity and improve the making them more difficult to use. Both use and comparison of nationally collected methods identify about a quarter of the U.S. data. These topologies usually are based on population as rural or “nonmetropolitan,” but the following features: population size and these populations are not identical. For ex- density; urbanization; adjacency and rela- ample, about 40 percent of the Census- tionship to an MSA; and principal economic defined rural population live within MSAs, activity. Although a standard typology may and 14 percent of the MSA population live in be desirable, it will be difficult to arrive at, Census-defined rural areas. The Census’ because the different topologies have merit rural population includes residents of small for various purposes. Nevertheless, there towns and cities but excludes those living in continues to be a need for a standardized towns larger than 2,500, many of whom nonmetropolitan topology. It is especially might be considered rural. MSAs can include important to display vital and health statistics areas that are sparsely populated and could be in a standardized way, because markedly dif- considered rural, while nonmetropolitan areas ferent conclusions can be reached, depending show significant within-area variation. on the definition of rural used. Better measures of population concentration or dis- There is no uniformity in how rural persion within counties would be helpful-- areas are defined for purposes of Federal especially for sparsely settled “frontier” areas program administration or distribution of --to distinguish between urban and rural funds. Different designations may be used areas within the same counties. 2. INTRODUCTION

Although there has been widespread con- tent differences, most notably lower popula- cern regarding a “health care crisis” in rural tion densities (1 3). By 1980, however, over areas, there is little agreement as to what two-thirds of the work force both inside and rural areas are. How rural areas (or rural outside of metropolitan areas were employed populations) are defined is far from academ- in three industries--service, manufacturing, ic, since urban/rural designations are basic to and retail trade (49). participation in certain Federal programs and to payment rates from Federal sources. In- The purpose of this staff paper is to: deed, the perceived magnitude of rural health care problems and the impact of any change in public policy depend on how rural is 1. describe the principal “rural” definitions defined. applied by the Federal Government that affect health programs and policies -- The features most intuitively associated i.e., urban and rural areas (and popula- with rurality are small populations, sparse tions) as defined by the Bureau of the settlement, and remoteness or distance from Census and metropolitan statistical areas large urban settlements. Historically, rural as defined by the Office of Manage- populations have been distinguished from ur- ment and Budget; ban ones by their dependence on farming oc- 2. describe the classifications used to dis- cupations and by differences in family size, tinguish different types of rural areas; lifestyle, and politics ( 13). However, because 3. discuss how Federal agencies have used of dramatic improvements in transportation these definitions to compile vital and and communication, migration to and from health statistics and to implement pro- rural areas, and diversification of the rural grams; and economy, these clear distinctions no longer 4. discuss the strengths and weaknesses of exist. The presence of farms, mining areas, rural definitions and classifications cur- and forests in rural areas contribute to persis- rently in use.

3 3. DELINEATING “RURAL” AND “URBAN” AREAS

The concepts of “rural” and “urban” now ther into farm (see below) and nonfarm pop- exist as part of a continuum. While few ulations. would argue about the extremes of that continuum- - e.g., an isolated farming com- Urbanized areas consist of a central core munity in Texas at one extreme and New (a “central city or cities”) and the contiguous, York City at the other--where to draw the closely settled territory outside the city’s line between urban and rural has become political boundaries (the “urban fringe”) that more difficult. Many Federal policies, how- combined have a total population of at least ever, rely on dichotomous rural/urban desig- 50,000 (48). The boundary of an urbanized nations. This section describes the two most area is based primarily on a residential popu- important dichotomous geographic designa- lation density of at least 1,000 persons per tions: the Bureau of the Census’ urban and square mile (the area generally also includes rural areas (and populations), and the Office less densely settled areas, such as industrial of Management and Budget’s (O MB) parks) (49). The boundaries of urbanized metropolitan statistical areas and residual areas are not limited to preexisting county or nonmetropolitan territory. Several geographic State lines; rather they often follow the classification schemes are then described that boundaries of small Census-defined geog- portray the urban-rural continuum. raphic units such as census tracts and U.S. Bureau of the Census enumeration districts. Many urbanized areas cross county and/or State lines (see figure 1). According to the Census Bureau, urban and rural are “type-of-area concepts rather than specific areas outlined on maps” (50). Figure 1--- Urbanized Areas The urban population includes persons living Urbanized area in urbanized areas (see below) and those living in places with 2,500 residents or more outside of urbanized areas. The population not classified as urban comprise the rural population; i.e., those living outside of urbanized areas in “places” with less than 2,500 residents and those living outside of “places” in the open countryside. Census-recognized “places” are either: 1) incorporated places such as cities, boroughs, towns, and villages; or 2) closely settled population centers that are outside of urbanized areas, do not have corporate limits, and have a population of at least 1 B County D 1,000. The rural population is divided fur- /

SOURCE : U.S. Department of Comnerce, Bureau of the Census, IICenSUS and Geography-Concepts and The minimum population of these unincorporated 1 Products, 08 Factf i rider CFF No. 8 (Wash- areas, called census designated places, is lower in ington, DC: U.S. Government Printing Of- Alaska and Hawaii. fice, August 1985).

5 6 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

Table 1--- Urban and Rural Population by Size of Place (1980)

Number Percent of places Population of Us.

U.S. total 22,529 226,545,805 100.0% Urban areas 8,765 167,050,992 73.7 Places of 1,000,000 or more 6 17,530,248 7.7 Places of 500,000-999,999 16 10,834,121 4.8 Places of 250,000-499,999 34 12,157,578 5.4 Places of 100,000-249,999 117 17,015,074 7.5 Places of 50,000-99,999 290 19,786,487 8.7 Places of 25,000-49,999 675 23,435,654 10.3 Places of 10,000-24,999 1,765 27,644,903 12.2 Places of 5,000-9,999 2,181 15,356,137 6.8 Places of 2,500-4,999 2,665 9,367,826 4.1 Places of less than 2,500 1,016 1,260,246 0.6 Other urban areaa 12,662,718 5.6 Rural areas 13,764 59,494,813 26.3 Places of 1,000-2,499 4,434 7,037,840 3.1 Places under 1,000 9,330 3,863,470 1.7 b Other rural area 48,593,503 21.4

~Includes urban residents not living in Census-designated places. Includes rural residents not living in Census-designated places and residents of the rural portion of extended cities. SOURCE: 1980 Census of Population, Volume 1, Characteristics of the Population, 1981, table 5, p. 1-37.

The 1980 Census identified 373 urbanized persons per square mile) in the immediate vi- areas in the United States and Puerto Rico cinity of cities of 50,000 or more population (52).2 were counted as urban instead of rural begin- ning in 1950 (21). With the exclusion of The Census definition of urban areas has these suburban residents, the size of the 1950 changed considerably over time. Prior to rural population dropped from 62 million to 1900, the lower population limit for the size 54 million (47). of places considered urban was set at either 4,000 or 8,000. The limit was lowered to The rural population has been divided by 2,500 residents in 1900(47). This definition the Census Bureau into the farm and nonfarm worked well until suburban development out- populations. The farm population includes side corporate boundaries became extensive. people living in rural-areas on properties of 1 To improve the definition, people living in acre of land or more where $1,000 or more of fairly densely populated areas (at least 1,000 agricultural products were sold (or would have been sold) during the previous 12 months.3 In 1987, the farm population was

z Since 1970, rural areas have been recognized ~ithin certain cities whose corporate limits include large areas lacking urban development. The rural portion of these Ilextended cities~ is at 3 From 1960 to the mid 1970s, the farm population least 5 square miles in area and has a population consisted of all persons living in rural territory density of less than 100 persons per square mile. on places of 10 or more acres, if at least $50 Together, such areas must constitute at least 25 worth of agricultural products were sold frcnn the percent of the land area of the legal city or in- place during the preceding 12 months. Persons living on places of under 10 acres were also include at least 25 square miles (50). In 1980 there were 87 extended cities with a total of 161,140 cluded if agricultural sales totaled $250 or more rural residents (41). (55). Defining “Rural” Areas: Impact on Health Care Policy and Research • 7

Table 2. --Ten States With The Largest Rural ceeds the 1,456 population of Hayward. The Population (1980) residents of the surrounding area use Hayward’s facilities such as a nursing home Rural population Percent and fire station but are not included in the State (in 1,000s) of State village population. This “undercount” has hampered the village’s ability to obtain grants Pennsylvania a 3,643 30.7 North Carolina 3,059 52.0 to improve area services (13). Numerous Texas 2,896 20.4 areas such as Hayward, that are considered Ohio 2,879 26.7 “rural” by virtue of the fact that they are out- Michigan 2,711 29.3 New York 2,700 15.4 side of an urbanized area and have a popula- California 2,060 8.7 tion of 2,500 or less, would be considered ur- Georgia 2,054 37.6 ban if the population immediately surround- Indiana 1,965 35.8 Illinois 1,908 16.7 ing the corporate area were included. Many towns and villages have resolved this problem SOURCE: U.S. Department of Commerce, Bureau of the by annexing surrounding developed territory Census, County and City Data Book: 1983 (12). (Washington, DC: U.S. Government Printing Office, 1983). Table 3--- States With More Than One-Half of Their Population Residing estimated at 4,986,000, or about 8 percent of in Rural Areas (1980) the rural population and 2 percent of the total resident U.S. population. In contrast, Rural population Percent farm residents represented 30 percent of the State (in 1,000s) of State population in 1920(55). Vermont 339 66.2 West Virginia 1,244 63.8 According to the 1980 Census, 73.7 per- South Dakota 370 53.6 cent of the U.S. population was urban, but Mississippi 1,328 52.7 the proportion ranged from a low of 33.8 Maine 591 52.5 North Carolina 3,059 52.0 percent in Vermont to 100 percent in the North Dakota 334 51.2 District of Columbia (51). Table 1 shows the distribution of the 1980 urban and rural pop- SOURCE: U.S. Department of Commerce, Bureau of the Census, Count Y and City Data Book: 1983, ulation by size of place. Over 85 percent of (Washington, DC: U.S. Government Printing the rural population live in places or areas Office, 1983). with fewer than 1,000 residents. Table 2 shows the ten States with the largest rural populations. Table 3 shows the seven States The Office of Management and Budget: with more than one-half of their population Metropolitan Statistical Areas residing in rural areas. A metropolitan statistical! area (MSA)4 is The Census Bureau’s ‘urbanized” area an economically and socially integrated geo- concept does not apply to towns, cities, or graphic unit centered on a large urban area. population concentrations of less than 50,000. In general terms, an MSA includes a large Those living nearby, but outside of the limits population center and adjacent communities of smaller cities or towns are not counted as that have a high degree of economic and so- being part of an “urbanized” area, even though the “suburban” population may be large and economically integrated with the town. For example, the population surround- 4 From 1959 to 1983, HSAS uere cal[ed Standard ing the incorporated village of Hayward, Wis- Metropolitan Statistical Areas (SMSAS) (53 FR 51175). The term MSA is used throughout this consin (county seat of Sawyer County), ex- paper, even when referring to 1980 Census data. 8 • Defining “Rural” Areas: Impact on Health Care Policy and Research

cial integration with that center (54). This often include more than one county; i.e., one contrasts with Census’ urban area, which is or more central counties containing the area’s defined solely on the basis of where people main population concentration and outlying reside (i. e., population size and density). counties that have close economic and social MSAs are defined by OMB5 and are used by relationships with those central counties. To Federal agencies for collecting, tabulating, be included in the MSA, the outlying coun- and publishing statistical data. Some Federal ties must have a specified level of commuting agencies also use MSA designations to imple- to the central counties and must also meet ment programs and allocate resources al- certain standards regarding metropolitan though OMB does not define them with such character, such as population density (see ap- applications in mind. The business com- pendix A). Consolidated MSAs(CMSAs) are munity uses MSA data and rankings ex- large metropolitan complexes within which tensively, for example to make investment individual components are defined, desig- decisions and to assess the desirability of nated as primary MSAs (PMSAs) (see appen- markets (38). dix A).

The official standards that are used to Problems in MSA classification may oc- define MSAs are reviewed prior to each cur when county boundaries do not conform decennial Census. 6 According to standards closely to actual urban or suburban develop- adopted for the 1980 Census, an MSA must ment. An MSA may inappropriately include have: 7 nonsuburban areas located in the outlying sections of some counties. For example, in a ■ a city with 50,000 or more residents; or spatially large county with a concentrated ■ an urbanized area (as defined by the metropolitan area, a large, sparsely populated Census Bureau) with at least 50,000 area maybe included in the MSA. This people that is part of a county or problem occurs more frequently in the West, counties that have at least 100,000 where counties are bigger than those in the people. East. On the other hand, an MSA may exclude suburban areas just across the county In most areas, counties are the building line. For example, a county with a suburban blocks of MSAs. In the six New England population that commutes to a neighboring States, MSAs are composed of cities and MSA may be excluded from that MSA be- towns, rather than whole counties.8 MSAs cause it also includes a large, sparsely populated section and therefore has a low average population density. 9 While these problems s The metropolitan area concept appeared in U.S. occur, they occur infrequently (56). Census publications as early as 1910 but Has not widely incorporated or used until the 1950 census when the concept was generalized to county lines About three-quarters (76.6 percent) of (12,47) . the U.S. population lived in the 275 MSAs designated as of 1983.10 These MSAs 6 The Office of Management and Budget’s Statistical Pol icy Office, Office of Information and Regulatory represent only 16.2 percent of the total U.S. Affairs, reviews and revises MSAS with advice from the interagency Federal Executive Committee on Metropolitan Statistical Areas (56).

7 See appendix A for a summary of the 1980 MSA 9 See appendix B for a description of criteria used standards. in including outlying counties in an USA.

8 New England USA standards are based primarily on 10 By June 30, 1988, intercensal population population density and commuting patterns (56). estimates or special census population counts had The six New England States are Maine, New Hamp- been used to add seven newly qualified HSAS and to shire, Vermont, Massachusetts, Rhode Island, and designate three new central cities within existing Connecticut. MSAS (12). Defining “Rural” Areas: Impact on Health Care Policy and Research • 9

10 • Defining “Rural” Areas: Impact on Health Care Policy and Research

Table 4--- Ten States With The Largest Table 5--- States With More Than One-Half Nonmetropolitan Population (1986) of Their Population Residing in Non- metropolitan Areas (1986) Nonmetropolitan Percent State population (in 1,000s) of State Nonmetropolitan Percent State population (in 1,000s) of State Texas 3,209 19.2 North Carolina 2,847 45.0 Idaho 80.7 Ohio 2,277 21.2 Vermont 416 76.9 Georgia 2,182 35.7 Montana 619 75.6 Illinois 2,033 17.6 South Dakota 508 71.8 Kentucky 2,033 54.5 Wyoming 361 71.2 Mississippi 1,837 70.0 Mississippi 1,837 70.0 Pennsylvania a 1,830 15.4 Maine 63.9 Michigan 1,811 19.8 West Virginia 1,217 63.4 Indiana 1,760 32.0 North Dakota 426 62.7 Arkansas 1,439 60.7 SOURCE : U.S. Bureau of the Census, Statistical Iowa 1,629 57.1 Abstract of the United States: 1988, 108th Alaska 56.0 ed. (Washington, DC: 1987), table 33. Kentucky 2,033 54.5 Nebraska 848 53.1 New Mexico 776 52.5 land area (figure 2.--MSA map). Seventy- SOURCE: U.S. Bureau of the Census, Statistical Ab- seven percent of U.S. counties (2,422 of 3,139 stract of the United States: 1988, 108th counties and county equivalents) are non- ed. (Washington, DC: 1987), table 33. metropolitan. 11 Table 4 shows the 10 States with the largest nonmetropolitan populations. Table 5 shows the 15 States with more than with metropolitan areas, such as hospitals one-half of their population residing in non- with comprehensive services, a 4-year col- metropolitan areas. lege, a local bus service, a TV station, or a Sunday paper (6). Before 1970, an MSA’s “recognized large population nucleus” had to include a central A few counties that have not qualified city of at least 50,000 population or twin for MSA status on the basis of demographic cities with a total population this large. Now characteristics have become designated as there is no minimum population size for an MSAs through the Federal legislative process. MSA’s central city, and it is easier to include Specifically, since 1983, one new MSA contiguous populations in the urbanized area (Decatur, Alabama) has been created (com- 12 (6). With the relaxation of MSA criteria, prising two counties) and the boundaries of two existing MSAs have been enlarged by some of the 58 MSAs designated following 13 the 1970 and 1980 censuses are demographi- statute (62). The proponents of the bill to cally dissimilar from those MSAs meeting create the Decatur, Alabama MSA argued that earlier standards. For example, of the 33 “MSA status would encourage a measure of MSAs newly designated after the 1980 census economic recovery to this area... without any that lacked a city of 50,000 or more resi- additional financial burden on the Federal dents, 25 had rural population percentages Government” (45). Hospitals located in the that were closer to nonmetropolitan norms (62 newly designated MSA of Decatur, Alabama percent) than metropolitan norms (15 percent) are expected to receive an additional $3 mil- (6). Furthermore, many of these do not have lion per year in Medicare reimbursements be- facilities and services traditionally associated

IZ Public Lau 100-258.

11 There were 717 metropolitan comties (excluding 13 Public Law 100-202, Sec. 530 and Ne~ England) as of June 30, 1988 (12). PWlic Law 99-500. Defining “Rural” Areas: Impact on Health Care Policy and Research Ž 11 cause of this change from nonmetropolitan Table 6--- Selected Federal Depart- (rural) to metropolitan status. The increase in ment/Agencies Using MSA Designations for Medicare outlays for these two counties the Administration of Programs or the Dis- would in aggregate decrease reimbursement to tribution of Fundsa other hospitals because the total amount of funding for the Medicare program was not Department of Agriculture changed by this act (44). Farmers Home Administration Rural Housing Assistance Department of Education The MSA definition is designed strictly Higher Education Assistance for statistical applications and not as a Federal Impact Payments for Education general-purpose geographic framework. In Sumner Food Service Program fact, according to official standards, “no Fed- Department of Health and Human Services Federal Grants for Residency Training eral department or agency should adopt these Aid to Organ Procurement Organizations statistical definitions for a nonstatistical pro- Medicare Prospective Payment System gram unless the agency head has determined Juvenile Delinquency Treatment Grants Provision of Services to Medicare Beneficiaries that this is an appropriate use of the classifi- by Health Maintenance Organizations (HMOs) cation” (56). The OMB does not take into ac- Department of Housing and Urban Development count or attempt to anticipate any nonstatisti- Enterprise Zones cal uses that may be made of the MSA Public Housing Development Community Development Block Grant Program definitions and will not modify the defini- Urban Development Action Grants tions to meet the requirements of any non- Assisted Housing Fair Market Rents statistical program (62). Nonetheless, Federal Rental Rehabilitation Awards agencies often use MSA designations to im- Department of the Interior Recreation Areas plement their programs. Table 6 contains a Wastewater Treatment Works Grants partial list of Federal programs that use Department of Labor MSAs for the administration of programs or Job Training Partnership Act the distribution of funds. %ost USA applications listed Mere identified by searching the U.S. Code and the Code of Federal Regulations (CFR) for the term 1’MSA. IC This list is not caqwehensive. SOURCE: Bea, K., llMetrwlitan Statistical Area Standards: Applications in Federal Policy, w (CRS Draft), 1989; U.S. Depart- ment of Commerce, OFSPS, “Report on the Inpct of Standard Metropolitan Statisti- cal Areas on Federal Programs,ll 1978. 4. RELATIONSHIP BETWEEN URBAN/RURAL AND METROPOLITAN/NONMETROPOLITAN DESIGNATIONS

Conceptually, the urban/rural and metro/ incorrectly classify as urban small towns nonmetropolitan designations are quite dif- which are located far from a large population ferent. Urban/rural are geographic designa- center. In contrast, the “nonurbanized area” tions based on population size and residential definition includes as rural all territory out- population densities, while the MSA concept side of its densely populated area, regardless embodies both a physical element (a city and of population size. Thus, while all “rural its built-up suburbs) and a functional dimen- areas” would be included, some cities and sion (a more-or-less unified local labor towns of as large as 40,000 residents would market) (21 ). The Census-defined urban also be included, as well as some outer sub- population and the MSA population intersect urbs of large urban areas. but are by no means identical; they are even less congruent geographically. Common to Figure 3--- The Relationship Between both are residents of most urbanized areas, Metropolitan Statistical Areas (MSAs), the densely settled area that forms the Urbanized Areas, and Urban and Rural Areas nucleus of the MSA (see figure 3).1 The Census’ urban population includes the urbanized area population and those living outside urbanized areas in places with 2,500 or more residents. The MSA population generally includes all those living in the county or counties that contain the urbanized area and the residents of additional counties that are economically integrated with that metropolitan core. Forty percent of the 1980 rural population lived in MSAs, and 14 percent of the MSA population lived in rural areas (see table 7). About one-fourth of farm residents ● live in MSAs (55).

“Rural area,” “nonurbanized area, ” and ● “nonmetropolitan area” have all been used to Urban places display vital and health statistics or to implement Federal policies in health and other • areas. These “rural” definitions can be ana- Rural areas lyzed in terms of how well they include “rural areas” and how well they exclude “urban areas.” The Census-defined “rural area” is the most specific measure, since it excludes urbanized areas and places with 2,500 residents Counties 1 through 4 comprise the MSA. or more. Thus, few would argue that an area Urbanized areas form the nucleus of the MSA and can span tuo or more counties (e. g., counties 1 through designated as rural according to the Census 4). There are a few urbanized areas i n non-MSA definition is really urban. However, some counties (e. g., county 7). might argue that the Census definition would Urban areas include urbanized areas and pl aces (e. g., cities and towns) with 2,500 or more residents. Such places are called urban places. Rural places are located outside of urbanized areas and have fewer than 2,500 residents. 1 There are a few urbanized areas outside of MSAs. Rural areas are the residential territory (shaded gray) left after urbanized areas and urban places z A sma[ 1 number of rural residents of extended are excluded. The USA has rural areas within it. c i t i es are exc [ uded from the urban and urbanized area population. SOURCE : Office of Technology Assessment, 1989. 13 14 • Defining “Rural” Areas: Impact on Health Care Policy and Research

Figure 4--- Map of California Counties: San Bernardino County

SOURCE : American Map Corporation, Business Control Atlas 1988 (Maspeth, New York: American Map Corpora- tion, 1988). - Defining ‘Rural” Areas: Impact on Health Care Policy and Research Ž 15

The nonMSA designation falls in between Table 7 .--Population Inside and Outside of the other two designations. If nonMSAs are MSAs by Urban and Rural Residence (1980) used to define rural areas, some large towns 3 Percent of and cities located outside of MSAS would be Population MSA/nonMSA included as rural while small towns and sparsely populated areas within MSAs would be U.S. total 226,545,805 excluded from the rural category. This ex- Inside MSAs 149,430,623 100.0 clusion is less a concern in the Eastern Urban 145,442,528 85.8 United States, where counties are relatively Urbanized areas 137,481,718 81.1 4 Central cities 66,222,207 39.1 small, and such towns would generally be Urban fringe 71,259,511 42.1 expected to be relatively close to an ur- Rural 23,988,095 14.2 banized area. However, in some of the large Outside MSAs 57,115,182 100.0 counties in the West, some areas within an Urban 21,608,464 37.8 Rural 35,506,718 62.2 MSA are far from an urbanized area (e.g., San Bernardino County--figure 4). SOURCE: U.S. Department of Commerce, Bureau of the Census, 1980 Census of Population, Volume 1. Characteristics of the Population, 1981, table 6, pp. 1-39.

3 There are at least 100 places with populations of 25,000 or more wtside of MSAS. d A typical county in the East has a land area of 400 to 600 square miles. Uest of the Mississippi River there are great variations, but the average county land area is just over 1400 square miles excluding Alaska (29). 5. UNDERSTANDING DIVERSITY WITHIN RURAL AREAS: URBAN/RURAL TOPOLOGIES

Dichotomous measures of urbanity/ search purposes and have not yet been used rurality not only obscure important dif- by Federal agencies to implement health ferences between urban and rural areas but policies or to present vital and health also wide variations within rural areas. Con- statistics. Before discussing specific sequently, there have been recommendations topologies, four geographic/demographic to implement a standard rural typology that measures common to most of the topologies would capture the elements of rural diversity are briefly described: 1 ) population size, 2) and improve use and comparison of data (14). population density, 3) adjacency to In the absence of such standardized data, it is metropolitan area, and 4) urbanization. difficult to quantify rural health problems and to make informed policy decisions. Population Size. --Population size can refer to the total population of the county or In this section, several county-based to the largest settlement in the county. rural/urban topologies or classification Presentation of an area’s population by settle- schemes are described that incorporate one or ment size helps to illustrate how the popula- more of the following measures: tion is distributed. In 1980, 43 percent of the U.S. population lived in places of less than ■ population size and density; 10,000 population or the open countryside ■ proximity to and relationship with urban (see table 1). The Census Bureau’s urban areas; definition depends in part on population size ■ degree of urbanization; and (i.e., those living in places of 2,500 or more principal economic activity. outside of urbanized areas). Only county-based topologies are consid- Population Density. --Population density ered here, because the county is generally the is calculated by dividing the resident popula- smallest geographic unit for which data are tion of a geographic unit by its land area available nationally. Counties also have measured in square miles or square kilo- several other characteristics that make them meters. In 1980, half of the U.S. population useful units of analysis: county boundaries are (excluding Alaska and Hawaii) lived in generally stable; counties can be aggregated counties with less than 383 persons per square up to the State level; and counties are impor- mile (21 ). Population density ranges from tant administrative units for health and other 64,395 persons per square mile in New York programs. For small-area analyses and for County, New York (Manhattan) to 0.1 per research purposes, ZIPCodes may be useful square mile in Dillingham Census Division,2 units of analysis. However, ZIPCodes bound- Alaska. Figure 5 shows how the U.S. popula- aries are not stable and sometimes cross tion is distributed. Urbanized areas are county lines. defined primarily by population density (i.e., territory with at least 1,000 residents per Topologies Used To Describe square mile). One drawback of population Nonmetropolitan Areas density is that it doesn’t describe how the Several topologies have been developed population is distributed within an area. For to classify nonmetropolitan counties. Nine example, a spatially large county that includes county-based topologies are described below.1 both small, densely settled urban areas and These topologies are generally used for re- large, sparsely populated areas would have a population density that masks such extremes.

1 Not al 1 rural topologies that have been proposed are described i n this sect i on. Excluded from dis- z There are no counties in Alaska. The county cussi on are severa 1 economic indices developed i n equi vat ents are the organized boroughs and O1census the 1960s that associated economic underdevelopnent areas]’ (U. S. Dept. of Commerce, 1980 Census of ui th rural i ty. Population, Volune 1, 1981). 17 18 • Defining “Rural” Areas: Impact on Health Care Policy and Research Defining “Rural” Areas: Impact on Health Care Policy and Research ■ 19

Adjacency to Metropolitan Area. --A Table 8--- Classification of Nonmetropolitan county’s adjacency to a metropolitan area can Counties by Urbanization and Proximity be measured geographically (e.g., sharing a to Metropolitan Areas boundary) or functionally (e.g., proportion of (2,490 counties as of 1970)’ residents commuting to an MSA for work). Many residents of these adjacent counties, Urbanized adjacent (173 counties) however, live some distance from an urban ■ Counties with an urban population of at least 20,000 which are adjacent to a metropolitan center, particularly in large counties in the county. West. Furthermore, natural geographic bar- Urbanized nonadjacent (154 counties) riers or an absence of roads may impede ac- ■ Counties with an urban population of at least cess to metropolitan areas. 20,000 which are not adjacent to a metropolitan county. Leas urbanized adjacent (565 counties) Urbanization --- Some topologies use vari- ■ Counties with an urban population of 2.500 to ous measures of the level of urbanization to 19,999 which are adjacent to a metropolitan differentiate nonmetropolitan counties. count y. Sometimes, urbanization is measured by the Less urbanized nonadjacent (734 counties) Counties with an urban population of 2,500 to absolute or relative size of the Census- 19,999 which are not adjacent to a metropolitan defined urban population. For non- county. metropolitan counties this generally means the Rural adjacent (241 counties) population living in places with 2,500 or Counties with no places of 2,500 or more popula- more residents or proportion of the county’s tion which are adjacent to a metropolitan county. Rural nonadjacent (623 counties) population that is urban. In other topologies, ■ Counties with no places of 2,500 or more popu- an urbanized county is defined by the size of lation which are not adjacent to a metropolitan the county’s total population (e.g., counties count y. with 25,000 or more residents). a Classification of nonmetropolitan areas using 1980 Census data is forthcoming from the Department of Urbanization/Adjacency to Agriculture (McGranahan, personal communication , Metropolitan areas 1989) . SOURCE : McGranahan et al., 1986, “Social and Eco- Analysts at the U.S. Department of Agri- nomic Characteristics of the Population in culture (USDA ) have classified non- Metro and Nonmetro Counties, 1970-1980.11 metropolitan counties on two dimensions: 1 ) the aggregate size of their urban population than a single point) and by commuting pat- and 2) proximity/adjacency to metropolitan 3 terns (i.e., at least 1 percent of the county’s counties (see table 8) (22). The urban popu- labor force commutes to the central lation follows the Census Bureau’s definition. 4 county(ies) of the MSA) . Nearly 40 percent Urbanized counties are distinguished from of the nonmetropolitan counties are adjacent less urbanized counties by the size of the ur- to MSAs, and just over one-half of the non- ban population (i.e., urbanized counties have metropolitan population resides in these ad- at least 20,000 urban residents and less ur- jacent counties (see table 9). banized counties have 2,500 to 19,999 urban residents). A nonmetropolitan county’s adjacency to an MSA is defined both by shared boundaries (i.e., touching an MSA at more d The classification scheme Has introduced in 1975 by Hines, Brown , and Zimner of USDA. Calvin Beale and David Broun, also at USDA, later modified the 3 This classification also includes three types of classification to include the 1 percent comnu t ing met ropo 1 i t an count i es based on HSA tot a 1 requirement for adjacent counties (13). A 2 per- popu 1 at i on- - sma 11 (under 250,000 popu 1 at ion), me- cent consnuting leve[ is used in a more recent ver- dium (250,000 to 999,999), and large ( 1 mi 1 I ion or sion of the typology (5). more). 20 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

This typology still masks differences of a large town or city does not guarantee among nonMSA counties. For example, both easy access to facilities for all residents of a a county with one town of 20,000 and a spatially large county. county with eight towns of 2,500 would be considered urbanized under this typology. Population Density: Incorporation of the The county with several small towns is un- Frontier Concept likely to have the level of services of a county with its population concentrated into The National Rural Health Association larger towns. (NRHA) has proposed a classification system that includes four types of rural areas (27) Adjacency to Metropolitan Areas/Largest adjacent rural areas--counties contiguous Settlement Size to or within MSAs which are very similar to their urban neighbors; Another county typology groups nonurbanized rural areas--counties with metropolitan counties by adjacency to MSAs 25,000 or more residents but distant and by size of the largest settlement (21) from an MSA; (table 10). Size of largest settlement is a use- frontier areas-- counties with population ful parameter to include when analyzing densities of less than 6 persons per health services since large settlements are square mile, which are the most remote more likely to have hospitals and specialized areas; health care providers. However, the presence

Table 9-- Nonmetropolitan County Table IO--- U.S. Population by County’s Population Distribution by Degree of Largest Settlement and Adjacency Urbanization and Adjacency to an MSA to an MSA (1980) (1980) Population Percent (1,000s) of Us. a b Population Percent ( 1,000s) of nonMSA U.S. total 226,505 100.0

U.S. total 226,546 NonMSA counties 60,512 26.7 Counties not adjacent to an USA MSA counties 163,526 Largest settlement Under 2,500 4,543 2.0 NonMSA counties 63,020 100.0% 2,500 to 9,999 10,255 4.5 Urbanized 10,000 to 24,999 7,120 3.1 Adjacent to MSA 14,802 23.5 25,000 or more 4,124 1.8 Not adjacent to MSA 9,594 15.2 Counties adjacent to an MSA Largest settlement Less urbanized Under 2,500 3,157 1.4 Adjacent to MSA 15,350 24.4 2,500 to 9,999 13,236 5.8 Not adjacent to MSA 15,529 24.6 10,000 to 24,999 12,467 5.5 25,000 or more 5,610 2.5 Totally rural MSA counties 165,994 73.3 Adjacent to MSA 2,737 4.3 Largest settlement Not adjacent to MSA 5,008 7.9 Under 100,000 3,611 1.6 100,000 to 249,999 18,461 8.2 a Total MSA/nonMSA populations differ from those in 250,000 to 499,999 24,883 11.0 table 7 because this typlogy relies on 1970 MSA 500,000 to 999,999 28,640 12.6 designations. 1,000,000 to 2,999 ,999 50,524 22.3 b 3,000,000 or more 39,875 17.6 Percent does not sun to 100 due to rounding.

SOURCE: D. A., McGranahan, et al. , “Social and SOURCE: Adapted from L., Long, and D., DeAre, llRepopu(ating the countryside: A 1980

Economic Characteristics of the Popula- M tion in Metro and Nonmetro Counties, Census Trend, Science, vol. 217, Sept. 1970-1980.11 17, 1982, pp. 111-116. Defining “Rural” Areas: Impact on Health Care Policy and Research • 21

• countryside rural areas--the remainder Table 11-- Bluestone and Clifton County of the country not covered by other Classifications Based on Urbanization and rural designations. Population Density

This typology includes some important Population per square concepts not covered by other topologies, Percent urban mile such as the concept of the “frontier” area. This typo logy also differs from other Bluestone classification topologies because it includes some counties Metropolitan GT 85 percent GT 100 within MSAs (i.e., in the adjacent rural area category ). Since the categories are not GT 50 percent GT 500 mutually exclusive, however, some counties Urban LT 85 percent 100-500 will fall into more than one group. For ex- Semi - isolated urban GT 50 percent LT 100 ample, under this typology 3 of 14 counties Densely settled rural LT 50 percent 50-100 in Arizona would be both “urbanized rural Sparsely settled rural LT SO percent LT 50 with some areas” and “frontier areas” because the urban population counties’ populations exceed 25,000 residents Sparsely settled rural O percent LT 50 and the population density is less than 6 per- with no urban sons per square mile. 5 County population size population is a poor indicator in the West because many Clifton's classification counties there are much larger than else- Urban GE 50 percent GE 200 where. Semi - urban GE 50 percent 30-200 Densely settled rural LT 50 percent GT 30 Urbanization/Population Density Rural LT 100 LT 30

Two other rural topologies incorporate ABBREVIATIONS: GT=greater than; GE=greater than or population density and urbanization. The equal to; LT=less than. first is a classification developed by SOURCE: B., Sinclair, and L., Manderscheid, “A 6 Comparative Evaluation of Indexes of Bluestone and the second is a modification Rurality--Their Policy Implications and by Clifton of that classification (see table Distributional Impacts, t’ contract report, 11).7 Urbanization is defined in terms of the Department of Agricultural Economics. proportion of the county that is urban (i.e., lives in towns of 2,500 or more). An ad- sures of urbanization and density provides vantage of using the percent of a county’s some indication of the degree of population population that is urban is that it is not in- concentration or dispersion. However, as fluenced much by the size of the county, or with the USDA typology, a county with one by a county’s including a large stretch of un- town of 20,000 and a county with eight towns populated territory. Density is heavily af- of 2,500 may not be distinguished under this fected by these conditions. Combining mea- scheme.

Distance From an MSA or Population Center 5 The three Arizona counties are Apache, Coconino, and Mohave. Two rural indexes8 are based on distance 6 Herman Bluestone, II FOCUS for Area Deve 10pment from an MSA or population center. Hathaway Ana [ ys is: Urban Or ientat ion of Counties, Economic et al., developed a size-distance index that Development Di vision, Economic Research Service, USDA as ci ted in Sinclair and Nanderscheid.

7 I very Clifton, Agricultural Economist, Economic 8 These rural indexes are different frun topologies Research Service, USDA, unpubl i shed manuscript as in that they are continuous (e. g., a scale fr~ 1 ci ted by Sinclair and Manderscheid. to 100) rather than categorical measures. 22 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research includes two measures: miles from an MSA a good access indicator and population size and the population of that MSA (39). Smith indicates service availability. The topologies and Parvin considered three county charac- incorporating these measures may be most in- teristics in their rural index: population- formative for geographically small counties. proximity; population density; and employ- For large counties, however, the distance ment in agriculture, forestry, or fisheries from one county seat to the next is unlikely (40,43). A county’s population-proximity in- to be applicable to those living at a distance dicates the relative access to adjacent from the county seat. counties’ populations. Commuting-Employment Patterns Population-proximity is measured as the county population plus the size-distance ratio A relatively new county classification of surrounding counties.9 To illustrate, the system incorporates measures of population population-proximity for County A of size size, urbanization, commuting patterns of 20,000 surrounded by four counties B workers, and the relationships between work- through E is as follows: place and place of residence (28). The classification criteria are shown in table 13 and the Table 12--- Population-Proximity: A Measure distribution of U.S. counties according to this of a County’s Relative Access to Adjacent typology is shown in table 14. The inclusion Counties’ Populations of employment and commuting measures may allow this typology to identify groups of Distance between Ratio of counties that are economically related such as County A and the population service and labor market areas. indicated county to distance count y population (miles)a (pop./mile) Economic and Socio-Demographic A 20,000 Characteristics B 15,000 30 50: c 60,000 1,500 D 250,000 100 2,500 Nonmetropolitan counties have also been E 100,000 10 10,000 classified according to their major economic bases, land uses, or population characteristics Sun of ratios ...... 14,500 10 Add population of County A...... 20,000 (table 15) (7). Fifteen percent of non- Population-proximity for County A...... 34,500 metropolitan counties (370 of 2,443 counties in the 48 conterminous States) remain un- aDistance is the number of miles between the county classified using this approach. Among the seat of County A and the county seat of the indicated county. counties that are classified, 70 percent fall into only one of the seven categories; the SOURCE : Adapted from Select Committee on Aging, 1983 "Status of the Rural Elderly." remaining 30 percent fall into two or more categories (37).

The combination of distance to adjacent Some of the data used to develop this population centers and size of that population classification are now a decade old (e.g., farm in a typology is attractive because distance is employment), and it is likely that with con- tinued diversification of the rural economy

9 The population-proximity is “the sun of the total population in the reference comty and the sun of the ratios of the ntmber of persons in all counties within 125 miles of the reference county divided by the distance in miles bet~een the county seat in the reference county and the county seat in each 10 These represent the nonmetropolitan counties as county within the specified distance (43).U defined in 1974. Defining “Rural” Areas: Impact on Health Care Policy and Research ■ 23

I I I 1 I I I I

I I i I 1

I I I

I I I I I I I I I I I i I I I I

I I I I I I I I I I I I I I I I I I t I I I I I I I I I I I I I I I I 1 I I I I

1 I I 1 I I I I I I I I I I I I I I

I I I I I I

I I 1

I 1 I I I I

I I I 1 1 I I

I I I I I

I I I I I I I I I I I I I I I I I I I I I I I I t I I i I I I I I I I I I I w I I I o I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I : I I t I I I I I 1 I I I I 1 I I I I I 1 I I 1 I I I I I I I I I I I I I I I I I I I I I I I I I I 24 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

Table 14--- Distribution of U.S. Counties by topologies are varied and require inclusion of Typology Based on Employment, Commuting, different measures. For example, to study and Population Characteristics (1986) the geographic variation of access to health care, a typology that includes population size, Number of Percent density, and distance to large settlements is of count i es of Us. interest. To study health personnel labor market areas, however, a typology based on Nonmetropolitan county trees 2393 23.2 economic areas, market areas, or worker Centers 543 11.1 commuting patterns is preferable. On the Satellites 212 2.4 other hand, rural economists or sociologists commuting counties with center 239 2.7 Small centers 565 3.7 may be more interested in identifying Rural commuting counties 333 1.7 counties with economies dependent on farm- Rural counties 501 1.6 ing, mining, or forestry. Metropolitan county type 745 76.8 While no one typology meets all potential Metro centers 295 44.7 needs, there are several desirable features of Metro satellites 91 10.0 Metro commuting satellites 193 15.0 any typology. For example, for many pur- Metro suburban 133 6.6 poses it is helpful to have topologies with Metro dormitory 33 less than 1 mutually exclusive (i.e., nonoverlapping) categories. The National Rural Health Associa- SOURCE: J., Pickard, "An Economic Development County Classification for the United tion’s typology includes frontier (less than 6 States and its Appalachian County Types," persons per square mile) and urbanized rural Appalachian Regional Commission, Wash- counties (population of 25,000 or more and ington, DC June 1988. not adjacent to an MSA). Yet it is possible for counties to meet both criteria. since the late 1970s, even fewer counties11 The concept of urbanization is incor- would be classified into one of these groups. porated into several of the topologies. In On the other hand, many rural economies some cases, urbanization is determined by the remain small and dependent on a single in- absolute or relative size of a county’s urban dustry or occupation despite the economic population and in others, by the size of a diversification(7). county’s largest settlement. When the size of the urban population is used, a county with Conclusion one large city with the balance of the county sparsely populated, would be indistinguishable In summary, several topologies for non- from a county with several smaller towns. As metropolitan counties have been developed level of resources are likely to be city-size incorporating measures of population size and dependent, topologies using this measure of density, urbanization, adjacency and rela- urbanization may not discriminate well for tionship to MSA, and principal economic ac- some applications. On the other hand, while tivity (see table 16). While it is desirable to largest settlement size might be indicative of have a standardized typology to portray the level of services available in the county, it is diversity of rural areas, the potential uses of not informative of how remote those services might be for all county residents. In geographically small counties, large settlements are likely to be accessible to all county residents. 11 If the classification scheme uere updated, the In the West, however, counties can be as proportion of nonmetropolitan counties either not classified or falling into more than one group large as some Eastern States, and some would likely be greater than the present 43 per- measure of proximity would be useful to in- cent. dicate physical access. Measures of how Defining “Rural” Areas: Impact on Health Care Policy and Research • 25

Table 15. --Classification of Nonmetropolitan Counties by Economic and Socio-Demographic Characteristicsa

Farming-dependent counties 702 counties concentrated largely in the Plains portion of the North Central region. Farming contributed a weighted annual average of 20 percent or more of total labor and proprietor income over the five years from 1975 to 1979. Manufacturing-dependent counties 678 counties concentrated in the Southeast. Manufacturing contributed 30 percent or more of total labor and proprietor income in 1979. MNining-dependent counties 200 counties concentrated in the West and in Appalachia. Mining contributed 20 percent or more to total labor and proprietor income in 1979. Specialized government counties 315 counties scattered throughout the country. Government activities contributed 25 percent or more to total labor and proprietor income in 1979. Persistent poverty counties 242 counties concentrated in the South, especially along the Mississippi Delta and in parts of Ap- palachia. Per capita family income in the county was in the lowest quintile in each of the years 1950, 1959, 1969, and 1979. Federal Lands counties 247 counties concentrated in the West. Federal land was 33 percent or more of the land area in a county in 1977. Destination retirement counties 515 counties concentrated in several northern Lake States as well as in the South and Southwest. For the 1970 to 1980 period, net immigration rates of people aged 60 and over were 15 percent or more of the expected 1980 population aged 60 and over. Retirement counties are disproportionately affected by entitlement programs benefiting the aged.

a The nutr&er of nonmetropolitan counties does not add to the total n-r (2,443), because the categories are not mutually exclusive and 370 counties do not fit any of the categories. SOURCE: Bender, L.D., Green, B.L., Hady, T.F., et al., Economic Research Service, U.S. Department of Ag- riculture, The Diverse Social and Economic Structure of Nomnetrocmlitan America, Rural Develop- ment Research Report No. 49 (Washington, DC: U.S. Goverrmnent Printing Office, Septenber 1985).

Table 16--- Features of the Nine County-Based Topologies

Measures Population Typology size Density Urbanization Adjacency Distance Economy

USDA-la . . . . ■ . . . . Long and DeAreb ■ . . . . ■ . . . . NRHAC ■ ■ ...... Bluestone d . . ■ ■ ...... Clifton e . . ■ ...... Parvin and Smithf ■ ...... Hathawayg ■ ...... ■ . . Pickard h ■ . . ■ . . . . ■ USDA-2i ...... ■

~cGranahan, D.A. et al., USDA, 1986. Long, L. and DeAre, D., 1982. ~~~~~~~\~ural Health Association, as cited i pat~~ L., ,989 . , H. as cited in Sinclair, B., and Mandersch&id,n L.V., 1974. ~C[ifton, I. as cited in Sinclair, B., and Manderscheid, L.V., 1974. Parvin, D.H. and Smith, B.J. as cited in U.S Congress, Hwse of Representatives, Task on the Rural Eldelry of the Select Comnittee on Aging, 1983. ‘Hathaway, D.E. as Cited in Sinc[air, B., and Manderscheid, L.V., 1974. ‘Pickard, J., Appalachia 21(3):19-24, Sumner, 1988. IBender, L.D. et al., USDA, 1985. SOURCE: Office of Technology Assessment, 1989. 26 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research evenly the population is distributed might A new category of nonmetropolitan area also be useful for large counties.12 Several of called “micropolitan area” has recently been the topologies incorporate an adjacent-to- described (42a). While not a typology, the MSA measure, which is an indicator of access new category does distinguish non- to level of services. The proportion of a metropolitan areas that exert similar social county’s population that is urban is a useful and economic influences on their regions as measure in large Western counties because metropolitan areas do on a larger scale. Most unlike population density, it is a measure that micropolitan areas are single counties but a is not influenced much by size of county or few span two counties or are independent by population distribution. cities. Micropolitan counties are relatively large (40,000 or more residents) and include a

Nonmetropolitan county data can also be central14,15 “core city” with at least 15,000 resi- disaggregate regionally by State or groups of dents.. Many micropolitan areas are COl- States (e.g., the four Census regions or nine lege towns, sites of military bases, and retire- Census divisions), or by economic areas (e.g., ment areas. More than 15 million people or Bureau of Economic Analysis Areas or about one-quarter of nonmetropolitan resi- BEAs). The Bureau of the Census defines dents live in the 219 identified micropolitan16 “county groups” that are usually contiguous areas. counties that combined have a population of 100,000 or more. 13 These counties are generally grouped according to meaningful State regions such as planning districts (50).

12 The Hoover index is a measure of population concentration or dispersion. The index ranges from 14 If a nonmetropol itan city of 15,000 or more zero, which indicates a perfectly uniform distrib- residents has at least 40 percent of its population ut ion in which each subarea has the same proportion in each of two counties, the micropol itan area in- of total population as it does of land area, to cludes both counties. 100, which represents the concentration of all the population into a single subarea (21). To estimate 15 In four States (Maryland, Missouri, Nevada, and county population dispersion, subcounty geographic Virginia) some cities (called independent cities) areas would be used. Other methods to measure have the sane status as counties and are considered population concentration or dispersion include the micropolitan if they have 15,000 or more residents nearest-neighbor statistic or the quadrant techni- and are larger than 15 square miles. If the city que, but both require a geographic information sys- is areally smaller, it is joined uith the adjacent tem incorporating longitude and latitude measures county to form the area. (9, 17,24) . 16 A list of micropolitan areas is available from ]3 These county groups are only defined in public Niagara Concepts, P.O. Box 296, Tonauanda, New York use data fi les. 14151”0296. 6. THE AVAILABILITY OF VITAL AND HEALTH STATISTICS FOR NONMETROPOLITAN AREAS

Given the diversity of nonmetropolitan the population aged 65 and older according to areas, it is important to present vital and metro/nonmetropolitan and urban/rural health and statistics by State, region, or by designations. The elderly appear to make up nonmetropolitan typology. Data from the a larger proportion of the total population in decennial Census and national vital statistics nonmetropolitan than metropolitan areas (13.0 (e.g., natality and mortality data) are pub- v. 10.7 percent). Using the urban/rural cate- lished for nonmetropolitan areas by State and gories, however, the opposite is true--there is degree of urbanization, but few other sources a greater proportion of elderly residents in of health information are published along urban than rural areas (11.4 v. 10.9). The these dimensions. For example, the National explanation of this discrepancy appears to be Center for Health Statistics does not publish that there are proportionately more persons detailed nonmetropolitan data (e.g., cross- 65 and older living in urban nonmetropolitan tabulated by Federal region) in their reports areas (14.3 percent) and fewer in rural on National Health Interview and National metropolitan areas (9.0 percent). Moreover, Medical Care Utilization and Expenditure when nonmetropolitan county MSA-adjacency Surveys. Sometimes, limitations of the way and size of the urbanized population are con- in which the data are collected (e. g., the sidered, the aged appear to be over- sample size or frame) limit the extent to represented in the less urbanized and non- which nonmetropolitan data can be displayed. adjacent counties (see table 18). In general, however, survey data files are available for public use and can be analyzed Table 18--- Proportion of Nonmetropolitan by area. Population Age 65 and Older by Level of Urbanization and Adjacency The choice of definition of “rural” used to an MSAa (1980)b to present demographic and health data can make a substantive difference. For example, Percent whether a disproportionate number of rural U.S. Population age 65 residents are elderly depends on how rural is ( 1,000s) and older defined. Table 17 shows the proportion of U.S. total 226,546 11.2 Metropolitan counties 163,526 10.7 Nonmetropolitan counties 63,020 12.8 Table 17--- Proportion of the Population 65 Urbanized and Older by Metropolitan/Non metropolitan Adjacent to metro area 14,802 11.9 and Urban/Rural Residence Not adjacent 9,594 11.0 Less urbanized Percent age Adjacent to intro area 15,350 13.3 Area U.S. population 65 and over Not adjacent 15,529 13.5

Metropolitan 169,430,577 10.7 Totally rural Nonmetropolitan 57,115,228 13.0 Adjacent to metro area 2,737 13.7 Urban 167,054,638 11.4 Not adjacent 5,008 14.6 Rural 59,491,167 10.9 ~rbanized counties are those with an urban popu- Metropolitan lation of at least 20,000; less urbanized counties Urban 145,451,315 10.9 are those with an urban population of between Central cities 67,854,918 11.8 2,500 to 19,999; and totally rural counties are Not central cities 77,596,397 10.2 those with no populations of 2,500 or more. Rural 23,979,262 9.0 Nonmetropolitan b1980 Census information is displayed using the Urban 21,603,323 14.3 1970 classification of counties. Rural 35,511,905 12.2 SOURCE: D. A., McGranahan, et al. , “Social and Economic Characteristics of the Popula- SOURCE : U.S. Department of Commerce, Bureau of the tion in Hetro and Nonmetro Counties, Census, 1980 Census: General Social and 1970-80, 11 USDA, ERS, Rural Development Economic Characteristics. Research report 58, appendix, table 2.

27 28 • Defining “Rural” Areas: Impact on Health Care Policy and Research

Infant mortality is also better understood Table 19.--Nonmetropolitan Infant Mortality by looking beyond metropolitan/nonmetro- Rates by Urban Area and Race, U.S. Total politan comparisons. Department of Health and Alabama (1986) and Human Services (DHHS) publishes data on infant mortality for urban and “not urban” Infant mortality rate (no. deaths) places within metropolitan and non- (deaths under age 1 per 1,000 births) metropolitan counties (nonmetropolitan urban United places are defined as those with populations States Alabama of 10,000 or more). 1 Table 19 shows that within U.S. nonmetropolitan areas ( 1985- Nonmetropolitan 10.4 (17,926) 12.7 (553) Urban placesa 10.8 (4,075) 10.9 (115) 1986), white infant mortality rates were lower White 9.9 (3,019) 7.4 (47) in nonurban places than in urban places (9.3 Black 16.5 (958) 16.3 (67) versus 9.9). Black infant mortality, in con- Other 7.1 (98) 7.6 (1) trast, is higher in non urban places (17.8 Balance of area 10.3 (13,851) 13.3 (438) White 9.3 (10,644) 10.5 (228) versus 16.5). In some nonmetropolitan areas Black 17.8 (2,632) 19.2 (210) (e.g., Alabama), infant mortality is higher in Other 10.7 (575) . . (o) the more rural areas for both whites and blacks (see table 19). aUrban places in nonMSA counties are those with populations of 10,000 or more. In summary, quite different conclusions SOURCE : Department of Health and Human Services, Public Health Service, Vital Statistics of about the rural population may be reached by the U. S.: 1986, 1985, Vol. 1, Natal ity, changing the definition of rural areas. Fur- Pub. No. 88-1123, 88-1113 (Washington, DC: thermore, important within-area variations U.S. Government Printing Office, 1988, 1987); 1986, 1985, Vol. 2, Mortality, Pub. are obscured when national data are not pub- No. 88-1114, 88-1102 (Washington, DC: U.S. lished for sub nonmetropolitan areas. Govermnent Printing Office, 1988, 1987). The problem of limited rural data is not a new one for policy makers. In 1981, the recommended that data be organized in a National Academy of Sciences addressed the building-block approach so that different issue in a report, Rural America in Passage: definitions and topologies could be con- Statistics for Policy. A panel on Statistics for structed. The panel recognized the need for Rural Development Policy comprised of agri- a common aggregation scheme for counties. cultural economists, statisticians, geographers, It recommended the development of a stan- sociologists, and demographers made a num- dard classification of nonmetropolitan ber of recommendations to improve the per- counties related to the level of urbanization. ceived poor availability and quality of rural The panel recommended that if possible, the statistical databases. The panel recommended county classification should be supplemented that the Federal Government “take a more ac- by a distinction between urban and rural tive role in the coordination of statistical ac- areas within counties (13). tivities and in developing and promulgating The lack of consistent county coding common definitions and other statistical stan- poses difficulties for those interested in de- dards that are appropriate for implementation veloping county-based definitions and at the Federal, State, and local levels.” The topologies. Unique county identifiers called panel concluded that a single definition of county FIPS (Federal Information Processing “rural” is neither feasible nor desirable but Standards) codes are provided by the National Institute of Measurement and Technology

1 DHHS defines urban places in USA counties as those with populations of 10,000 or more but less than 50,000. This urban definition differs from z The National Institute of Measurement and Tech- the Bureau of the Census definitions of urban or nology was formerly the Bureau of National Stan- urbanized areas. dards. Defining “Rural” Areas: Impact on Health Care Policy and Research Ž 29 but are not universally used (8). The panel preferentially locating in the outlying sub- recommended that Federal and State data be urban areas of MSAs. recorded with such county codes to permit tabulations for individual counties and groups Maps effectively illustrate geographic of counties. Adherence to a county coding variation in health status and access to health system would facilitate aggregation of in- care resources. U.S. cancer atlases have been formation regardless of how rural is defined. published at the county level providing a Since the report was issued in 1981, few of visualization of geographic patterns of cancer mortality not apparent from tabular data its recommendations have been implemented 4 (8). (60). Rural women in the lower socio- economic classes have high rates of cervical The relative merits of the county-based cancer and for white women, maps show topologies for health service planning and re- concentrations of cervical cancer throughout search can be evaluated using the Area the South, especially in Appalachia (see fig- Resource File (ARF), a county-level data ure 6). base maintained by the Health Resources and Maps of the United States by county Services Administration (61). The file con- show higher death rates due to unintentional tains data necessary for the Bureau of Health injury (e.g., housefires and drownings) and Professions to carry out its mandated program motor vehicle crashes in rural areas, particu- of research and analysis of the geographic larly in Western, sparsely populated counties distribution and supply of health personnel. (see figures 7-8). The large volume of travel Population, economic, and mortality data, and on major routes traversing rural areas does measures of health personnel, health educa- not account for the high rural death rates. tion, and hospital resources, are included in Instead, road characteristics, travel speeds, the file (61). seat-belt use, types of vehicles, and The ARF has been used to show how the availability of emergency care are factors that availability of physician and hospital may contribute to the excess of motor vehicle resources varies by type of nonmetropolitan crash deaths in rural areas (3). area (table 20) ( 18). For example, when Maps of nonmetropolitan county varia- physician availability is examined by type- tion in health indicators (e. g., infant of-county, wide variations in physician-to- mortality) and the distribution of health care population ratios are evident. The average resources (e.g., physicians, hospitals) will soon physician-to-population ratio is 64 per 5 3 be published in the Rural Health Atlas. A 100,000 in nonmetropolitan counties but it typology of rural medical care is being devel- ranges from 131 per 100,000 in high-density oped for the Atlas, which incorporates counties to a low of 45 per 100,000 in persis- measures of access to primary care physicians tent poverty counties (see table 20). Some- and health facilities. what surprisingly, there appear to be relative- Such a typology will help identify isolated communities with ly more physicians in nonadjacent than ad- limited access to health care (35). jacent nonmetropolitan counties (67 compared to 59 per 100,000). A possible explanation is that physicians serving many of the residents d The U.S. Cancer Atlas maps cancer mortality by of the adjacent nonmetropolitan counties are county groupings called State Economic Areas (SEA). 506 SEAS were delineated by the Bureau of the Census in 1960. SEAS are geographic units ~ith sirni lar demographic, climatic, physiographic, and cultural features (60). 3 This analysis was ( imi ted to nonmetropolitan count i es of 1 ess than 50,000 population i n 1985. J The atlas is scheduled to be published by re- Only physicians engaged in patient care are in- searchers at the University of North Carolina by C 1 Uded. October, 1989 (35). 30 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

Table 20.--Characteristics of Different Categories of U.S. Nonmetropolitan Counties (2,092 nonmetropolitan counties of less than 50,000 population in 1985)8

1985 1986 1986 1980 1979 Category M.D.+ hospital hospital Age X in (numbet of counties) DO/100,000 beds/1,000 days per 1,000 over 65 poverty

U.S. total (2092) 64.2 5.0 962 14.2 17.6 Urbanized (83) 113.7 6.4 1421 12.5 15.2 Less urban (1239) 71.9 5.5 1081 13.9 16.7 Rural (770) 46.5 4.1 721 15.1 19.3

MSA adjacent (751) 58.6 4.3 858 13.9 16.4 MSA nonadjacent (1341) 67.3 5.4 1021 14.8 18.2

1980 population density 3 or less (194) 48.9 4.9 13.1 17.9 >3 and < 6 (181) 59.2 7.2 1382 14.7 16.5 >6 and < 9 (123) 63.4 6.1 1035 15.9 16.1 >9 and < 50 (1235) 60.5 4.6 858 14.8 18.5 >50 and < 100 (320) 80.5 4.9 1053 12.5 15.7 more than 100 (39) 130.5 7.7 1959 11.4 12.0

East (59) 115.7 5.5 1443 13.5 12.8 South Atlantic (324) 60.7 4.2 12.7 20.7 South (624) 54.4 4.3 14.8 22.0 Central (799) 64.9 5.9 1193 16.0 14.3 West (286) 75.4 5.1 942 11.5 14.3

Agricultural only (464) 52.2 5.7 1o11 16.6 17.1 Agricultural total (680) 49.1 5.1 944 15.9 18.8 Manufacturing only (290) 68.3 4.5 847 13.2 15.2 Manufacturing total (500) 62.4 4.3 824 13.4 16.8 Mining only (97) 61.2 5.1 774 12.2 16.0 Mining total (183) 57.1 4.3 689 11.8 16.5 Federal Lands only (35) 106.8 3.8 698 10.0 12.0 Federal lands total (210) 75.8 3.9 643 11.4 14.8 Government only (75) 76.5 9.9 2382 13.4 18.0 Government total (246) 66.6 7.0 1603 13.2 19.4 Poverty only (41) 45.3 3.4 535 13.5 29.9 Poverty total (238) 43.0 3.3 575 13.6 28.3 Retirement only (140) 79.1 4.5 841 16.9 16.0 Retirement total (420) 67.5 4.0 743 15.6 17.6 a 282 nomtetropolitan counties uith 50,000 or more population were excluded from analyses.

SOURCE: Kindig, D.A., et al., llN~tropo(itan County Typology and Health Resources;ti unpublished manuscript, Dec. 15, 1988. Defining “Rural” Areas: Impact on Health Care Policy and Research ■ 31

Figure 6--- Areas With Cervical Cancer Mortality Rates Significantly Higher Than the U.S. Rate, and in the Highest 1OO/o of all SEA Rates (White Females, 1970-1980)

SOURCE: U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, Atlas of U.S. Cancer Mortality Among Whites: 1950-1980, DHHS Pub. No. (NIH) 87-2900 (Bethesda, MD: 1987). 32 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

Figure 7--- Death Rates Due to Unintentional Injury by County

0.00 — I I 38.16 — 72.07 —

90.44 — 429.00

SOURCE : Baker, S. P., Uhitfield, R. A., and O]Nei 11, B., qwty Mapping of Injured Mortal ity, ” The Journal of Trauna 28(6):741-745, Jme 1988. Defining “Rural” Areas: Impact on Health Care Policy and Research • 33

Figure 8--- Death Rates Due to Motor Vehicle Crashes by County

57.28– 1465.20

SOURCE : Baker, S. P., Uhitfield, R.A., and O’Neill, B., llGeOgr@ic variations in Mortality From Motor Vehicle Crashes,U New England Journal of Medicine 316(22):1384-1387, May 28, 1987. 7. USING OMB AND CENSUS DESIGNATIONS TO IMPLEMENT HEALTH PROGRAMS

There is no uniformity in how rural because of differences in urban and rural areas are defined for purposes of Federal standardized amounts, average wage and program administration and distribution of case-mix indexes, and other factors. funds. Even within agencies different definitions maybe used. This may occur Rural hospitals designated as “sole com- when agencies implement programs or munity hospitals” are not subject to the same policies for which rural areas have been reimbursement methods as other rural hospi- defined legislatively. For example, the tals.3 These hospitals are “by reason of fac- MSA/nonMSA designations are used to cate- tors such as isolated location, weather condi- gorize hospitals as urban or rural areas for tions, travel conditions, or absence of other purposes of hospital reimbursement under hospitals, the sole source of inpatient hospital Medicare. On the other hand, in the case of services reasonably available in a geographic clinics certified under the Rural Health area to Medicare beneficiaries.” An excep- Clinics Act, “rural” is defined as Census tion is also made for large nonmetropolitan Bureau-designated nonurbanized areas. hospitals that serve as “rural referral centers” Certified clinics receive cost-based reim- for Medicare patients. These hospitals are bursement from Medicare and Medicaid. reimbursed at the same rate as urban hospitals These two examples of how the MSA and (58). Census designations are used are described in more detail in the following section. Finally, The rural/urban reimbursement differen- the definition of “frontier” areas is described tial has not been well-accepted by some hos- as it is used by the Department of Health and pitals. In some cases, the concerns of non- Human Services (DHHS). metropolitan hospitals have prompted legis- lators to change the designation of the county Medicare Reimbursement: Using MSAs To in which the hospital is located from non- Define Urban and Rural Areas metropolitan to metropolitan. The HCFA metropolitan/nonmetropolitan hospital reim- Several geographic designations affect bursement standards were modified by the hospital reimbursement under Medicare’s Omnibus Reconciliation Act of 1987.4 Some prospective payment system (PPS). Different hospitals located in nonMSAs were reassigned reimbursement rates are calculated for hospi- to the urban (MSA) category. Accordingly, a tals located in rural, large urban (population hospital located in a nonmetropolitan county of more than a million),l and other urban adjacent to one or more metropolitan area is areas. Under PPS, Congress directed the treated as being in the metropolitan area to Health Care Financing Administration which the greatest number of workers in the (HCFA) to define “rural” and “urban” hospi- county commute, if: tals as those located in nonmetropolitan and metropolitan areas, respectively.2 On aver- ■ the nonMSA county would otherwise be age, urban hospital per-case payments are 40 considered part of an MSA area but for percent higher than those of rural hospitals the fact that the nonMSA county does not meet the standard relating to the

1 I n Neu Engl and County Metropol i tan Areas (MECMAs), a large urban area includes a population 3 The prospective payment rates for sole cmnun i ty of more than 970,000. hospitals -al 75 percent of the hospital-specific base payment rate plus 25 percent of the ap- L Certain nornnetropol itan New England counties were propriate regional prospective payment rate (58). deemed to be parts of metropol i tan areas for purposes of PPS. 4 Pub(ic Law 100-203 Sec. 4005.

35 36 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

rate of commutation between the non the PPS formula should be refined so that MSA county and the central county or continuous measures are used to adjust a counties of any adjacent MSA; and single reimbursement rate. HCFA is examin- ■ either 1) the number of residents of the ing the feasibility of using severity measures non MSA county who commute for as a more sensitive alternative to geographi- employment to the central county or cally based separate rates (65). counties of any adjacent MSA is equal to at least 15 percent of the number of The Prospective Payment Assessment residents of the nonMSA county who are Commission (ProPAC), a body formed to employed; or 2) the sum of the number make recommendations to the Congress on of residents of the nonMSA county who PPS, has stated that before it can make a commute for employment to the central recommendation to either maintain or county or counties of any adjacent MSA eliminate separate urban and rural rates, it and the number of residents of any ad- must better understand why there is an ap- jacent MSA who commute for employ- proximate 40 percent difference in average ment to the nonMSA county is at least Medicare cost per case between urban and equal to 20 percent of the number of rural hospitals. This cost difference was residents of the nonMSA county who are present when the PPS rates were first estab- employed. lished and has persisted through at least the first three years of PPS. The PPS rural/urban Thirty-nine non MSA counties meet these payment differential reflects poorly un- standards (53 FR 38498). derstood geographic practice pattern variations that cannot be attributed to measurable Some hospitals dissatisfied with the differences in patient characteristics, quality rural/urban reimbursement differential have of care, or market area features. The issue is resorted to lawsuits in order to receive urban complicated by the unknown relationship be- rates. For example, 28 hospitals in Missouri tween practice pattern variations, revenues, nonMSAs have sued DHHS, contending that costs, and quality (34). MSA designations are not related to the costs of providing medical care and that DHHS Defining Rural Labor Market Areas --- underpays for the services provided to The PPS formula includes a wage index ad- Medicare patients. Under the current regula- justment that takes into account geographic tions, a hospital in Jefferson City, for exam- differences in labor costs. A different wage ple, is paid less than a hospital in Columbia index is applied to urban and rural labor 30 miles away, because the first hospital is market areas. Labor market areas are rather located outside an MSA ( 15). The National precisely defined for urban areas--each MSA Rural Health Association has filed a class ac- is defined as a labor market area. In con- tion suit against DHHS, charging that rural trast, there is one rural labor market area hospitals’ Fifth Amendment rights to due defined for each State, which includes all process are being violated on two counts re- nonMSA counties in that State. lated to “unreasonably low reimbursement for rural hospitals” (16). Recognizing wide variation in hospital wage levels within these broadly defined In a congressionally mandated study, labor markets, ProPAC has recommended that DHHS examined the feasibility and impact of rural hospital labor market areas be redefined phasing out or eliminating separate urban and to distinguish between urbanized rural rural payment rates, retaining regional or counties and other rural counties within each hospital-specific rates, refining the wage in- State. Accordingly, urbanized rural counties dex, and other alternatives to separate ur- would be defined as counties with a city or ban/rural rates (58). The study suggests that town having a population of 25,000 or great- Defining “Rural” Areas: Impact on Health Care Policy and Research • 37 er5 (33). Analyses of 1982 data show average shortage area (HMSA) or a medically un- hospital wages in State’s “urbanized rural deserved area (MUA). The practice must counties” to be 8.5 percent higher than wages use a mid-level practitioner (physician as- in “other rural counties” ($7.54 v. $6.95) (32). sistant or nurse practitioner) at least 60 per- DHHS asserts that wage differentials are al- cent of the time that the practice is open. ready taken into account to some degree There has been renewed interest in this Act through other PPS adjustments (i.e., the in- following an increase in the ceiling of rea- direct medical education and disproportionate sonable costs reimbursed by Medicare and share adjustments) and the special treatment Medicaid programs. The payment cap is in- for rural referral centers (53 FR 38498). dexed to the Medicare Economic Index (36).6

ProPAC has also recommended (31,32,33) Rural areas, for purposes of the Rural that definitions of urban hospital labor Health Clinics Act, are ‘areas not delineated market areas be modified to include a dis- as urbanized areas in the last census con- tinction between an MSA’s central urban and ducted by the Census Bureau.” Nonurbanized outlying areas. They suggest that urbanized areas encompass a larger area than either the areas within an MSA, as defined by the non MSA or Census-defined rural areas. Bureau of the Census, could be distinguished Therefore, Rural Health Clinics can be lo- from nonurbanized areas. DHHS has rejected cated within an MSA (see figure 3) or in a this proposal, in part because of the difficulty nonMSA town with a population of 2,500 or of assigning a hospital to an urbanized area, more (such a town is urban according to the the boundaries of which are defined below Census Bureau). the MSA level. Determining whether or not a hospital is inside or outside of an urbanized In summary, for purposes of hospital area involves pinpointing the hospital location reimbursement under Medicare, the MSA in terms of the smallest units of Census geog- designation is used (with certain specific ex- raphy (the block or block group). In a study ceptions) to distinguish urban from rural hos- conducted for ProPAC ( 1), a process is de- pitals. Persistent MSA/nonMSA hospital cost scribed whereby the location of a hospital can differences have been noted since the PPS be specified in terms of Census geography rates were first established, but it is likely and then mapped to urbanized area bound- that MSA location is an indirect measure of aries. According to DHHS, however, defin- hospital cost. Hospital-specific measures are ing labor markets below the county level being sought to replace the MSA adjustment would be confusing and difficult to ad- in the PPS formula. minister. Geographic designations are also used to The Rural Health Clinics Act define urban and rural labor market areas. Dissatisfaction with having only one rural Ambulatory services can be reimbursed labor market area per State (i.e., one labor on an at-cost basis by Medicare and Medicaid market for all non MSA counties) has led if facilities and providers meet certification ProPAC to recommend two labor market requirements of the Rural Health Clinics Act areas for nonMSA counties. They have sug- (Public Law 95-210). To be certified, a gested recognizing as urbanized, nonMSA practice must be located in a rural area that counties with a city or town with a popula- is designated either as a health manpower tion of 25,000 or greater (33). The average

5 This definition of an urbanized rural county 6 These changes to the Rural Health Clinics Act should not be confused u i th the Bureau of the uere contained i n the Budget Reconc i 1 i at i on Act of Census definition of an urban or urbanized area. 1987. 38 • Defining “Rural” Areas: Impact on Health Care Policy and Research hospital wage is 8.5 percent higher in ur- Health Center (CHC) grantees (and new ap- banized rural counties than in nonurbanized plicants for CHC support) and National rural counties (32). There are less than 125 Health Service Corps sites.8 Frontier areas nonMSA towns with 25,000 or more popula- were defined as (59): tion, so few of the 2,393 nonMSA counties Those areas located throughout the country would be classified as urbanized (49). In which are characterized by a small popu- fact, this distinction would create only 37 lation base (generally 6 persons per square new areas (32). mile or fewer) which is spread over a con- siderable geographic area. Although HCFA has chosen not to use urbanized areas to refine labor market areas, To be eligible for Bureau of Health Care HCFA does use urbanized area designations Delivery and Assistance (BHCDA) support as when certifying hospitals and clinics under a frontier area, the following service area the Rural Health Clinic Act. Rural Health criteria must be met (59):9 Clinics must be located in nonurbanized areas that are designated as either a health man- Service Area: a rational area in the fron- power shortage area or a medically un- tier will have at least 500 residents within deserved area. This liberal interpretation of a 25-mile radius of the health services “rural” (e. g., it includes some areas within delivery site or within the rationally estab- MSAs) seems appropriate, given the require- lished trade area. Most areas will have ment that the area must also be medically un- between 500 to 3,000 residents and cover deserved. This allows some medically un- large geographic areas. deserved areas within MSAs--but isolated from an urbanized area by factors other than Population Density: the service area will distance--to be certified. have six or fewer persons per square mile.

Providing Services in “Frontier” Areas Distance: the service area will be such that the distance from a primary care Health services may be difficult to pro- delivery site within the service area to the vide in large, sparsely populated areas. Areas next level of care will be more than 45 with a population density of 6 persons per miles and/or the average travel time more square mile or less, called “frontier” areas, are than 60 minutes. When defining the “next common West of the Mississippi river (30) level of care,” we are referring to a facil- (figure 9). In 1980, by this definition, there ity with 24-hour emergency care, with 24- were at least 378 frontier counties with a to- hour capability to handle an emergency tal population of nearly 3 million persons caesarean section or a patient having a (42). It may take an hour or more for resi- heart attack and some specialty mix to in- dents of frontier areas to reach health pro- clude at a minimum, obstetric, pediatric, viders and facilities. Frontier physicians tend internal medicine, and anesthesia services. to be generalists, solely responsible for a large service area, and have limited access to hospitals and health care technology (1 1). Recognizing the unique characteristics of 8 The 1988 authorizing legislation for Public frontier areas, 7 DHHS in early 1986 agreed to Health Service programs of assistance for prinmry health care included reconunendations for DHHS to use different criteria to evaluate Community support primary health care planning, development, and operations in frontier areas (46).

!) If the eligibility criteria are not strictly met, 7 The Frontier Task Force of the National Rural an organization may justify any unusual circum- Health Association (established in 1985) uas in- stances which may qualify them as frontier, for strumental in documenting the unique health care example, geography, exceptional economic condi- needs of rural areas (63). tions, or special health needs (59). Defining “Rural” Areas: Impact on Health Care Policy and Research • 39

Figure 9--- Frontier Counties: Population Density of 6 or Less

( + — ,~ A—-

SOURCE : U.S. Department of Health and Human Services, Public Health Service, Health Resources and Services Administration, Bureau of Health Professions, Office of Data and Management, Area Resource File, June 16, 1986. 40 • Defining ‘Rural” Areas: Impact on Health Care Policy and Research

Some State Health Departments have had It is useful to distinguish frontier area trouble identifying service areas meeting counties with evenly distributed small settle- these criteria (26). Whole counties can be ments from counties with one or two large identified as frontier areas on the basis of population settlements and large areas with population density, but available sub-county little or no settlement. For example, the geographic units are sometimes inadequate for health service needs of two frontier counties identifying health service areas. Population in New Mexico with similar population data from the 1980 Census are available for densities differ because of the way the popu- sub-county areas such as Census County lations are distributed. One county has a to- Divisions (CCDs), and Enumeration Districts tal population of approximately 8,000, of (EDs) (see appendix D) but these areas can be whom about 6,000 live in one town. In con- large and may not represent a rational health trast, the other county has a total population 11 - service area.10 ZIpCodes may be ag of 2,500 living in six widely dispersed towns. gregated to form a rational service area, but If suitable sub-county areas were available, this poses some technical difficulties (19). the Hoover Index, which measures population Following the 1990 Census, Block Numbering concentration or dispersion, could be used to Areas will be available for all nonurbanized distinguish between these counties. An auto- areas (see appendix D.--1980 Census geog- mated geographic information system called raphy).12 TIGER (Topologically Integrated Geographic

Encoding and13 Referencing System) has been0 developed that will enhance the ability ‘ 10 Some States have defined primary care service conduct spatial analyses of population data areas (e. g., Neu York). from the 1990 decennial census (23). 11 population data from the Census are available by Z I PCode. Some investigators have used ZIPCode- level census data to describe three types of rural area based upon density within zip code: semi-rural (density of 16 to 30 per square mile); rural (density 6 to 15 per square mile; and frontier (density less than 6 per square mile) (10).

12 In 1980, Block Nunbering Areas were only avail- 13 TIGER has been developed jointly by the U.S. able for nonurbanized places ~ith over 10,000 pop- Geological Survey and the U.S. Bureau of the ulation. census. 8. CONCLUSIONS

The concepts of “rural” and “urban” exist have been recommendations to implement a as part of a continuum, but Federal policies standard rural typology that would capture generally rely on dichotomous urban/rural the elements of rural diversity and improve differences based on designations of the Of- use and comparison of data. Nine county- fice of Management and Budget (OMB) or based rural/urban topologies or classification the Bureau of the Census. OMB’s MSA schemes that incorporate one or more of the designation includes a large population center following measures are reviewed in this and adjacent counties that have a high degree paper: population size and density; proximity of economic and social integration with that to and relationship with urban areas; degree center. Census’ urban areas include densely of urbanization; and principal economy. settled “urbanized areas” plus places with While a standard typology may seem desir- populations of 2,500 or more outside of ur- able, it will be difficult to arrive at, because banized areas. “Rural” areas are designated the different topologies are designed and by exclusion: i.e., those areas not classified as have merit for various purposes, some of either MSA or urban. About one-quarter of which conflict. the U.S. population resides in nonMSAs and Census’ rural areas. The identified popula- For purposes of health services planning tions are different but overlapping. Forty and research, a typology based on largest percent of the 1980 Census’ rural population settlement size is useful, because the level of lived in MSAs, and 14 percent of the MSA available health resources is likely to be re- population lived in rural areas. lated to the size of a city. In spatially small counties, large settlements are likely to be “Nonmetropolitan area,” “rural area,” and quite accessible to all county residents. In “nonurbanized area” have all been used to the West, however, counties can be several display vital and health statistics or to imple- times as large as in the East, and some ment Federal policies. These “rural” defini- measure of proximity would be useful. A tions can be analyzed in terms of how well measure of population concentration and dis- they include “rural areas” and how well they persion, or distance to a large settlement, exclude “urban areas.” For example, we in- could serve as an indicator of access to those tuitively associate farming with “rural” but services. Of the topologies reviewed in this about one-fourth of farm residents live in paper, the one likely to best measure both MSAs (55). Some might argue that isolated level of and access to services is a typology towns with just over 2,500 residents are in- that incorporates a county’s largest settlement appropriately excluded from the Census’ rural and the county’s adjacency to an MSA. definition. Others may argue that when non- Other topologies that categorize counties ac- MSAs are defined as rural, over 100 towns cording to employment and commuting pat- with populations of 25,000 or more are in- terns could be used to refine the definition of appropriately included. Moreover, when labor market areas, an important component MSAs are used to define “urban” in spatially of the Medicare prospective payment system large counties, small towns that are far from (PPS) formula. an urbanized area are inappropriately called urban. Rural areas are not defined uniformly for purposes of Federal program administra- Dichotomous measures of urbanity/ tion or distribution of funds. Different rurality obscure important differences be- designations may, in fact, be used by the tween urban and rural areas and wide varia- same agency. For example, Congress has tions within a rural area. Consequently, there directed the Health Care Financing Adminis-

41 42 • Defining “Rural” Areas: Impact on Health Care Policy and Research tration to use OMB’s MSA designations to large Western counties, sub-county geo- categorize hospitals as urban or rural for pur- graphic units need to be employed to help poses of hospital reimbursement under Medi- identify health service areas with special care, but to use Census’ nonurbanized area characteristics such as those that are ‘frontier” designation to certify health facilities under (i.e., have 6 or fewer persons per square the Rural Health Clinics Act. mile).

The relative merits of county-based The choice of definition for “rural” that topologies for particular applications can be is used to present demographic and health evaluated by using the Area Resource File data can make a substantive difference. For (ARF), a county-level data base maintained example, whether a disproportionate number by the Health Resources and Services Admin- of rural residents are elderly depends on how istration. In addition, visual aids such as rural is defined. Furthermore, wide varia- maps can effectively serve as an analytic tions in health status indicators within non- device to illustrate geographic variation in metropolitan areas will not be apparent unless health status and access to health care nonmetropolitan data are disaggregate by resources and could further the development region, urbanization, proximity to urban and evaluation of topologies. In the spatially areas, or other relevant factors. APPENDIX A: SUMMARY OF THE STANDARDS FOLLOWED IN ESTABLISHING METROPOLITAN STATISTICAL AREAS

This statement summarizes in nontechni- tory where the population density is less than cal language the official standards for desig- 1,000 persons per square mile. nating and defining metropolitan statistical areas. It omits certain exceptions and unusual Central County(ies) (Section 2) situations that are covered in the standards themselves or in the detailed statement of the Every metropolitan statistical area has procedures followed in applying the stan- one or more central counties. These are the dards. counties in which at least half the population lives in the Census Bureau urbanized area. Population Size Requirements for There are also a few counties classed as cen- Qualification (Section 1) tral even though less than half their population lives in the urbanized area because they To qualify for recognition as a metro- contain a central city (defined in Section 4), politan statistical area, an area must either or a significant portion (with at least 2,500 have a city with a population of at least population) of a central city. 50,000 within its corporate limits, or it must have a U.S. Bureau of the Census urbanized Outlying Counties (Section 3) area of at least 50,000 population, and a total metropolitan statistical area population of at In addition to the central county(ies), a least 100,000. A few metropolitan statistical metropolitan statistical area may include one areas that do not meet these requirements are or more outlying counties. Qualifications still recognized because they qualified in the an outlying county requires a significant level past under standards that were then in effect. of commuting from the outlying county to the central county(ies), and a specified degree The Census Bureau defines urbanized of “metropolitan character.” The specific re- areas according to specific criteria, designed quirements for including an outlying county to include the densely settled area around depend on the level of commuting of its resi- each large city. An urbanized area must have dent workers to the central county(ies), as a population of at least 50,000. The ur- follows: banized area criteria define a boundary based primarily on a population density of at least 1. Counties with a commuting rate of 50 1,000 persons per square mile, but also in- percent or more must have a population clude some less densely settled areas within density of at least 25 persons per square corporate limits, and such areas as industrial mile. parks, railroad yards, golf courses, and so 2. Counties with a commuting rate of 40 forth, if they are adjacent to dense urban de- to 50 percent can qualify if they have a velopment. The density level of 1,000 per- density of at least 35 persons per square sons per square mile corresponds approxi- mile. mately to the continuously built-up area 3. Counties with a commuting rate of 25 around the city, for example, as it would ap- to 40 percent typically qualify through pear in an aerial photograph. having either a density of at least 50 persons per square mile, or at least 35 Typically, the entire urbanized area is percent of their population classified as included within one metropolitan statistical urban by the Bureau of Census. area; however, the metropolitan statistical 4. Counties with a commuting rate of 15 area is usually much larger in areal extent to 25 percent must have a density of at than the urbanized area, and includes terri- least 50 persons per square mile, and in

43 44 Ž Defining “Rural” Areas: Impact on Health Care Policy and Research

addition must meet two of the follow- Consolidating or Combining Adjacent ing four requirements: Metropolitan Statistical Areas (Sections 5 and 6) ■ the population density must be at least 60 persons per square mile; These two sections specify certain condi- ■ at least 35 percent of the population must be classified as urban; tions under which adjacent metropolitan statistical areas defined by the preceding sec- ■ population growth between 1970 and tions are joined to form a single area. Sec- 1980 must be at least 20 percent; and tion 5 consolidates adjacent metropolitan ■ a significant portion of the popula- statistical areas if their commuting inter- tion (either 10 percent or at least change is at least 15 percent of the number 5,000 persons) must live within the of workers living in the smaller of the two urbanized area. areas. To be consolidated under Section 5, There are also a few outlying counties each of the metropolitan statistical areas must that qualify for inclusion in a metropolitan also be at least 60 percent urban, and the to- statistical area because of heavy commuting tal population of the consolidated metropol- from the central county (ies) to the outlying itan statistical area must be at least a million. county, or because of substantial total commuting to and from the central counties. Section 6 provides for combining as a single metropolitan statistical area those ad- Central Cities (Section 4) jacent metropolitan statistical areas whose largest cities are within 25 miles of each Every metropolitan statistical area has at other, unless there is strong evidence, sup- least one central city, which is usually its ported by local opinion, that they do not con- largest city. Smaller cities are also identified stitute a single area for general social and as central cities if they have at least 25,000 economic purposes. population and meet certain commuting requirements. Levels (Section 7)

In certain smaller metropolitan statistical This section classifies the prospective areas there are places between 15,000 and metropolitan statistical areas defined by the 25,000 population that also qualify as central preceding sections into four categories based cities, because they are at least one-third the on total population size: Level A with a mil- size of the metropolitan statistical area’s lion or more; Level B with 250,000 to a mil- largest city and meet commuting require- lion; Level C with 100,000 to 250,000; and ments. Level D with less than 100,000.

Most places that qualify as central cities Under this section, the metropolitan are legally incorporated cities. It is also pos- statistical areas in Levels B, C, and D (those sible for a town in the New England States, with a population of less than 1 million) New York, or Wisconsin, or a township in receive final designation as metropolitan Michigan, New Jersey, or Pennsylvania to statistical areas. qualify as a central city. The town or township must, however, be recognized by Area Titles (Section 8) the Bureau of the Census as a “census designated place” on the basis of being entirely ur- This section assigns titles to the metro- ban in character, and must also meet certain politan statistical areas defined by the population size and commuting requirements. preceding sections. Defining “Rural” Areas: Impact on Health Care Policy and Research ■ 45

Primary and Consolidated Metropolitan population density and commuting. Statistical Areas (Sections 9 through 11) For users who prefer definitions in terms Within the metropolitan statistical areas of counties, a set of New England County classified as Level A, some areas may qualify Metropolitan Areas is also officially defined. for separate recognition as primary metro- However, the official metropolitan statistical politan statistical areas. A primary metro- area designations in New England apply to politan statistical area is a large urbanized the city-and-town definitions. county, or cluster of counties, that demon- strates very strong internal economic and so- In order to determine the cities and cial links, in addition to close ties to the towns which could qualify for inclusion in a other portions of the Level A metro-politan New England metropolitan statistical area, statistical area. section 12 defines a central core for each New England urbanized area, consisting es- Section 9 through 11 provide a frame- sentially of cities and towns in which at least work for identifying primary metropolitan half the population lives in the urbanized statistical areas within metropolitan statistical area or in a contiguous urbanized area. areas of at least 1 million population. A metropolitan statistical area in which primary Once the central core has been defined, metropolitan statistical areas have been iden- Section 13 reviews the adjacent cities and tified is designated a consolidated metro- towns for possible inclusion in the metropolitan statistical area. politan statistical area. An adjacent city or town with a population density of at least 100 Metropolitan Statistical Areas persons per square mile is included if at least in New England (Sections 12 through 14) 15 percent of its resident workers commute to the central core. Towns with a density be- These sections provide the basic stan- tween 60 and 100 persons per square mile dards for defining metropolitan statistical also qualify if they have at least 30 percent areas in New England. commuting to the central core. However, the commuting to the central core from the city Qualification for recognition as a metro- or town must be greater than to any other politan statistical area in New England is on central core, and also greater than to any much the same basis as in the other States. A nonmetropolitan city or town. few modifications in the standards are necessary because cities and towns are used for the If a city or town has qualifying commut- definitions. In New England each Census ing in two different directions (e.g., to a cen- Bureau urbanized area of at least 50,000 tral core and to a nonmetropolitan city) and normally has a separate metropolitan statisti- the commuting percentages are within five cal area, provided there is a total metro- points of each other, local opinion is solicited politan statistical area population of at least through the appropriate congressional delega- 75,000 or a central city of at least 50,000. tion before assigning the city or town to a The total metropolitan statistical area popula- metropolitan statistical area. Some New Eng- tion requirement is lower than the 100,000 land communities also qualify for inclusion in required in the other States because the New a metropolitan statistical area on the basis of England cities and towns used in defining reverse commuting or total commuting. metropolitan statistical areas are much smaller in areal extent than the counties used for the Once the qualifying outlying towns and definitions in the other States. This makes it cities have been determined, Section 14 possible to define New England metropolitan qualifies the resulting area as a metropolitan statistical areas quite precisely on the basis of statistical area provided it has a city of at 46 • Defining “Rural” Areas: Impact on Health Care Policy and Research least 50,000 or a total population of at least metropolitan statistical areas. It follows the 75,000. This section also specifies that same general approach as is used for several of the standards used in the other identifying such areas outside New England States are also applied to the New England (Section 9). Finally, Section 16 provides that States: level and titles for New England primary and consolidated metropolitan statistical areas are 1. The central cities of each area are determined by much the same standards as determined by Section 4. for the remaining States. 2. Two adjacent New England metropolitan statistical areas may be consolidated under Section 5. 3. New England areas are categorized into levels according to Section 7A. Those in Levels B, C, and D are given final designation as metropolitan statistical areas, and are assigned titles according Note: OMB is reviewing the MSA standards to Section 8. and will publish them with some revisions before Apr. 1, 1990 (12). Primary and Consolidated Metropolitan Statistical Areas in New England SOURCE: Excerpt from “The Metropolitan (Sections 15 and 16) Statistical Area Classification: 1980 Official Standards and Re- Section 15 is used to review each Level lated Documents, ” The Federal A metropolitan statistical area in New Eng- Committee on Standard Metro- land for the possible identification of primary politan Statistical Areas. APPENDIX B: THE CENSUS BUREAU’S URBANIZED AREA DEFINITION

The major objective of the Census tion density of at least 1,000 persons Bureau in delineating urbanized areas is to per square mile.4 provide abetter separation of urban and rural population and housing in the vicinity of 3. Other contiguous unincorporated area large cities. An urbanized area consists of a with a density of less than 1,000 per- central city or cities and surrounding closely sons per square mile, provided that it: settled territory or “urban fringe.” ■ eliminates an enclave of less than 5 square miles which is surrounded by There are 366 urbanized areas delineated built-up area; in the United States for the 1980 census. ■ closes an indentation in the boundary There are seven urbanized areas delineated in of the densely settled area that is no Puerto Rico. more than 1 mile across the open end and encompasses no more than 5 The following criteria are used in square miles; and determining the eligibility and definition of 1 ■ links an outlying area of qualifying the 1980 urbanized areas. density, provided that the outlying area is: An urbanized area comprises an incorporated place2 and adjacent densely settled --connected by road to, and is not surrounding area that together have a mini- more than 1½ miles from, the main mum population of 50,000.3 The densely body of the urbanized area; and settled surrounding area consist of: --separated from the main body of the urbanized area by water or 1. Contiguous incorporated places or other undevelopable area, is con- census-designated places having: nected by road to the main body of the urbanized area, and is not ■ a population of 2,500 or more; or more than 5 miles from the main ■ a population of fewer than 2,500 but body of the urbanized area. having either a population density of 1,000 persons per square mile, closely 4. Large concentrations of nonresidential settled area containing a minimum of urban area (e.g., industrial parks, office 50 percent of the population, or a areas, and major airports), which have cluster of at least 100 housing units. at least one-quarter of their boundary contiguous to an urbanized area. 2. Contiguous unincorporated area which is connected by road and has a popula-

d Any area of extensive nonresidential urban land me, (e.g., railroad yards, airports, factories, ] All references to population counts and densities parks, golf courses, and cemeteries) is excluded in relate to data from the 1980 census. computing the population density. z In Hawaii, incorporated places do not exist in the sense of functioning local governmental units. Instead, census-designated places are used in defining a central city and for applying urbanized Note: The Census Bureau is reviewing the area criteria. urbanized area rules and will publish 3 The rural portions of extended cities, as defined them with some revisions by 1990. in the Census Bureauts extended city criteria, are excluded from the urbanized area. In addition, for SOURCE: Excerpt from 1980 Census of an urbanized area to be recognized, it must include Population Vol. 1, Characteristics a population of at least 25,000 that does not reside on a military base. of the Population, Appendix A. Area Classifications. 47 APPENDIX C: CENSUS GEOGRAPHY

CENSUS GEOGRAPHY-CONCEPTS INTRODUCTION

It is important for anyone using census data to be aware of the geographic Figure 1. CENSUS REGIONS AND GEOGRAPHIC DIVISIONS concepts involved in taking the census OF THE UNITED STATES and allocating the statistics to States, counties, cities, and smaller areas down to the size of a city block. Preparing for and taking a census also results in a number of geographic tools or products that are helpful to the data user as well as to the Census Bureau, in activities such as computerized location coding, mapping, and graphic display. They also allow users to interrelate local and census statistics for a variety of planning and administrative purposes. This Fact- finder explains the Census Bureau’s geographic concepts and products. Except where noted, the definitions and references below are those used for the 1980 Census of Population and Housing. Figure 10 on page 6 summarizes the geographic areas for which ● The Midwest Region was designated as the data are available from other Bureau North Central Region until June 1984. censuses and surveys. Data summarizes are presented in printed reports m , microfiche m, and computer tapes @ and flexible diskettes K , based on tabulations for the geographic and statistical levels discussed below. Maps El are also Figure 2. NATIONAL GEOGRAPHIC available. The symbols ● and +, keyed sions are groupings of States. (See RELATIONSHIPS fig. 1.) to the legend on page 3, indicate how to Nation obtain the items described in this • I brochure. Governmental units of the Nation— Regions States (50) and the District of Columbia Divisions REPORTING AREAS Counties and their equivalents (3,139, plus 78 in Puerto Rico) There are a number of basic relation- Minor civil divisions (MCD’s) of coun- ships, illustrated below, among the ties, such as towns and townships places’ designated geographic areas the Census Bureau places” (approximately 25,000) uses as “building blocks” in its reports. Incorporated places (about 19,100), Counties Some of the areas are governmental e.g., cities and villages Minor civil census unite, i.e., legally defined entities, while ● census county divisions (CCD’s)-In divisions county other areas are defined specifically for 20 States where MCD’s are not ade- divisions statistical purposes. (The statistical quate for reporting subcounty census / areas are italicized in the diagrams; all statistics, Bureau and local officials districts others are governmental.) delineated 5,512 CCD’s (plus 37 cen- or block groups sus subareas in Alaska) for this United States-The 60 States and the a Note that places (incorporated and census purpose. designated] are not shown within the county District of Columbia. (Data also are ● •Census designated places (CDP’s)– and county subdivision hierarchy, since collected separately for Puerto Rico Formerly referred to as “unincor- places may cross the boundaries of these areas. A few census reports and tape series and the outlying areas under U.S. porated places,” CDP’s (about 3,500) do show places within MCD or CCD within sovereignty or jurisdiction.) are closely settled population centers county, but in these cases data pertain only without legally established limits, to that part of a place which is within a Particular higher-level area. Enumeration Regions/divisions-There are four delineated with State and local district and block-group summaries do racog- Census regions defined for the United assistance for statistical purposes, nize place boundaries, making ED’s and BG’s important as the Iotwrst common de States, each composed of two or more and generally have a population of at nominator for the higher-lwel entities. geographic divisions. The nine divi- least 1,000.

48 —

Defining “Rural” Areas: Impact on Health Care Policy and Research Ž 49

● Census tracts –These statistical sub- Figure 3. GEOGRAPHIC Figure 4. URBAN/RURAL divisions of counties (approximately RELATIONSHIPS IN AN MSA GEOGRAPHIC RELATIONSHIPS 43,350, including 463 in Puerto Rico), MSA ALL AREAS average 4,000 inhabitants. They are / delineated (subject to Census Bureau Urban MetropolitanI standards) by local committees for /\ metropolitan areas and roughly 200 Urbanized Other Rural Rural other counties. ● Blocks–Generally bounded by streets and other physical features, Central Urban Rural Other blocks (approximately 2.5 million) are groups districts cities fringe places’ rural identified (numbered) in and adjacent I nonfarm Blocks a Includes both governmental and sta- to urbanized areas, most incor- tistical units. porated places of 10,000 or more a In New England, MSA’S are defined in population, and other areas that con- terms of towns and cities, rather than munties (as in the rest of the country), tracted with the Census Bureau to b Cen5us tr~ts subdivide moat MSA collect data at the block level. (Fig. counties as wel I as about 20I3 other cou ntiea, 8 illustrates the extent of block- As tracts may cross MCD and place bound- statistics coverage in part of a State.) aries, MCD’S and places are not shown in this hierarchy, Five States are completely block- numbered. other hierarchies; there can be both ● Block-numbering areas (BNA’s)– urban and rural territory within Areas (approximately 3,400, in- metropolitan as well as nonmetro- cluding over 100 in Puerto Rico) analyzed on the basis of the 1980 politan areas. defined for the purpose of grouping population census and a revised set of and numbering blocks where census criteria, these areas were redefined and There are other geographic units for tracts have not been established. the word “standard” was dropped from which data may be obtained from the ● Block groups (BG’s)–Subdivisions the titles. Thus, on June 30, 1983, 1980 Census of Population and Hous- of census tracts or BNA’s, BG’s SMSA’s and SCSA’s were redesignated ing. Some appear in regular publications (about 200,000) comprise all blocks as and data files: American Indian reser- with the same first digit in a tract or . Metropolitan statistical areas vations (278, both State and Federal, in- BNA. Averaging 900 population, (MSA'S) cluding 3 administered by or for more BG’s appear in areas with numbered ● Consolidated MSA's (CMSA's) than one tribe), Alaska Native villages blocks in lieu of ED’s (see below) for and (209), congressional districts (435), and tabulation purposes. ● Primary MSA (PMSA's) election precincts in some States. Data ● Enumeration districts (ED’s)–An As the 1982 Economic Censuses are prepared for neighborhoods in ED is a Bureau administrative area covered calendar year 1982, prior to the almost 1,300 areas and by ZIP Code assigned to one census enumerator. June 1983 date for adopting the areas nationwide. Data for other areas ED’s (about 100,000 nationwide) changes, the 1982 SMSA and SCSA are generated in special tabulations were used for census tabulation pur- designations and nomenclature were re- prepared at cost, for example, school poses where census blocks were not tained for those censuses. Some data districts. numbered. ED size varies con- from the 1980 Census of Population and Two types of areas are defined siderably, but averages 500 Housing were retabulated by MSA and specifically for the economic censuses: inhabitants. issued in special reports, and the new ● central business districts (CBD’s) Metropolitan Areas definitions were used in preparing population and migration estimates and CBD’s are areas of high land value, ● Standard metropolitan statistical traffic flow, and concentration of areas (SMSA‘s)-An SMSA (defined in presenting current statistics from 1983 onward. retail businesses, offices, theaters, by the Office of Management and hotels, and service establishments. In Budget) comprised one or more coun- ● Urbanized areas (UA's)–A UA (them the 1982 Census of Retail Trade, 456 ties around a central city or urbaniz- are 373, including 7 in Puerto Rico) CBD’s were defined in (1) any SMSA ed area with 50,000 or more in- consists of a central city and central city and (2) any other city habitants. Contiguous counties were surrounding densely settled territory with a population of 50,000 or more included if they had close social and with a combined population of 50,000 and a sufficient concentration of economic links with the area’s or more inhabitants. (See fig. 5) economic activity. CBD’s also are population nucleus. There were 323 ● Metropolitan/nonmetropolitan– shown in place -of-work data from the SMSA’s, including 4 in Puerto Rico. “Metropolitan” includes all popula- 1980 Census of Population and ● Standard consolidated statistical tion within MSA’s; “nonmetropoli- Housing. areas (SCSA‘s)--SCSA’s (17, in- tan” comprises everyone elsewhere. ● Major retail centers (MRC’s)-MRC’O cluding 1 in Puerto Rico) were com- ● Urban/rural-The urban population are concentrations of retail stores posed of two or more adjacent consists of all persons living in ur- located in SMSA’s, but outside the SMSA’s having a combined popula- banized areas and in places of 2,500 CBD’s. For 1982, 1,545 MRC’s were tion of 1 million or more, and with or more inhabitants outside these defined areas with at least 25 retail close social and economic links. areas. All other population is establishments and one or more large After the relationships between central classified as rural. The urban and general merchandise or department urban core(s) and adjacent counties were rural classification cuts across the stores. 50 • Defining “Rural” Areas: Impact on Health Care Policy and Research

Figure 6. GEOGRAPHIC HIERARCHY INSIDE AND OUTSIDE URBANIZED AREAS (UA’s) (See figures 7-10 for maps exhibiting most of these features.)

MSA

aThe entire MSA is subdlvlderl intO census tracts. bB10ck5 and biock 9rouPs do not have symbolized boundaries as do the other areas, but are identified by number. (see discu=ion on pa- 2.)

SOURCE : U.S. Department of Commerce, Bureau of the Census, "Census and Geography-Concepts and Products," Factfinder CFF No. 8 (Rev.) Washington, DC: U.S. Government Printing Office, August 1985). Defining “Rural” Areas: Impact on Health Care Policy and Research • 51

The Bureau collects and publishes data (MRCs) outside CBDs; in the Census of Gov- for two kinds of sub-state areas: ernments, for school districts and other special districts; and in foreign trade and inter- Governmental, such as-- national research, for countries and world incorporated places (e.g., cities, villages) areas. and minor civil divisions (MCDs) of counties (e.g., townships), Generally, survey data are published only congressional districts and election for the larger areas, such as the United precincts, and States, its regions, and some States, while American Indian reservations and Alaska census data are made available for smaller Native villages. areas as well. Statistical, including-- Population and Housing standard metropolitan statistical areas (SMSAs) and standard consolidated The decennial census of population and statistical area (SCSAs) were used in the housing is the most important source of data 1980 decennial and 1982 economic for small communities, not only on a wide censuses. In 1983, SMSAs and SCSAs variety of subjects but in finer geographic were replaced by metropolitan statistical detail than from any other statistical base. It areas (MSAs), primary MSAs (PMSAs), provides a uniform set of data for inter- and consolidated MSAs (CMSAs); community comparisons as well. census county divisions (CCDs) in States Table A-1 shows the items collected in where MCD boundaries are not satisfac- the census. The basic data, called “complete tory for statistical purposes; count” or “100 -percent,” come from the ques- census-designated places (formerly called tions asked for every person and housing “unincorporated places”); unit. Other items are obtained only at a urbanized areas; sample of households and housing units in census tracts (subdivisions of counties, order to keep response burden to a minimum. primarily in metropolitan areas) and block numbering areas (BNAs), averag- The 100-percent data provide the basic ing about 4,000 people each; population and housing counts and certain census blocks- -generally equivalent to characteristics -- e.g., age, sex, and race for city blocks in cities, but are very large people; and value or rent, and vacant or oc- in rural areas; cupied status for housing units--for all tabu- enumeration districts (EDs)--census ad- lation areas, even down to census blocks. ministrative areas, averaging around 700 Since they are estimates rather than complete inhabitants, used where block statistics counts, the sample statistics for small com- are not available; munities must be used with caution. block groups (BGs)--counterparts to EDs averaging 900 population, in areas with In general, the higher the geographic or census blocks; statistical level of tabulation, the greater neighborhoods --subareas locally defined amount of detail there is available in the by participants in the Bureau’s Neigh- census reports. With respect to small com- borhood Statistics Program; and munities, more data usually are contained in ZIPCodes--Postal Service administrative the printed reports at the county level than areas independent of either governmen- for the county subdivisions and places. (This tal or other statistical units. difference seldom occurs on summary tape files or selected microfiche). Only limited In the 1982 Census of Retail Trade, the county- and subcounty-level data are avail- Bureau published data for central business able on flexible diskettes and through districts (CBDs) and major retail centers CENDATA. 52 • Defining “Rural” Areas: Impact on Health Care Policy and Research

Table A-1. --Items Collected in the 1980 Census

100-percent population items 100-percent housing items Household relationship Number of housing units at address Sex Complete plumbing facilities Race Number of rooms in unit Age Tenure (whether the unit is owned or rented) Marital status Condominium identification Spanish/Hispanic origin or descenta Value of home (for owner-occupied units and condominiums) Rent (for renter-occupied units) Vacant for rent, for sale, etc., and period of vacancy Sample population items Sample housing items School enrollment Number of units in structure Education attainment Stories in building and presence of elevator State or foreign country of birth Year unit built Citizenship and year of immigration Year moved into this housea Current language and English proficiency Source of water Ancestry b Sewage disposal Place of residence 5 years ago Heating equipment Activity 5 years ago Fuels used for home heating, water-heating, and cooking Veteran status and period of service Costs of utilities and fuelsa Presence of disability or handicapa Complete kitchen facilities Children ever born Numbet of bedrooms and bathrooms Marital history Telephone Employment status last week Air conditioning Hours worked last week Number of automobiles Place of work Number of light trucks and vansb Travel time to workb Homeowner shelter costs for mortgage, a b Means of transportation to work real estate taxes, and hazard insurance Persons in carpool b Year last worked Industry Occupation Class of worker Amount of income by source in 1979a Work in 1979 and weeks looking for work in 1979a

~Changed relative to 1970. Neu item for 1980.

Derived items (illustrative examples)

Families Household size Family type and size Persons per room (“overcrowding") Family income Institutions and other group quarters Poverty status Farm residence Population density

Note: This information pertains to the 1980 census and does not reflect changes in data presentation and availability following the 1990 census.

SOURCE: Adapted from "Data for Small Communities,” U.S. Bureau of the Census--FACTFINDER for the Nation, CFF No. 22 (Rev.) January 1986. APPENDIX D: RURAL HEALTH CARE ADVISORY PANEL

James Bernstein, Panel Chair Director, Office of Health Resources Development North Carolina Department of Human Resources Robert Bergland T. Carter Melton, Jr. Executive Vice President and General Manager President National Rural Electric Rockingham Memorial Hospital Cooperative Association Harrisonburg, Virginia Washington, D.C. Jeffrey Merrill James Coleman Vice President Executive Director Robert Wood Johnson Foundation West Alabama Health Services, Inc. Princeton, New Jersey Sam Cordes Myrna Pickard Professor& Head Dean Department of Agricultural Economics School of Nursing University of Wyoming University of Texas - Arlington Elizabeth Dichter Carolyn Roberts Senior Vice President for Corporate Strategies President Lutheran Health Systems Copley Health Systems, Inc. Denver, Colorado Morrisville, Vermont Mary Ellis Roger Rosenblatt Director Professor& Vice Chairman Iowa Department of Public Health Department of Family Medicine University of Washington Kevin Fickenscher Assistant Dean and Executive Director Peter Sybinsky Michigan State University Deputy Director for Planning, Center for Medical Studies Legislation, and Operations Kalamazoo, Michigan Hawaii Department of Health Roland Gardner Fred Tinning President President Beaufort-Jasper (South Carolina) Kirksville College of Comprehensive Health Center Osteopathic Medicine Kirksville, Missouri Robert Graham Executive Vice President Robert Vraciu American Academy of Family Physicians Vice President Kansas City, Missouri Marketing& Planning HealthTrust, Inc. Alice Hersh Nashville, Tennessee Executive Director Foundation for Health Services Research Robert Walker Washington, D.C. Chairman Department of Family and David Kindig Community Health Director Marshall University School of Medicine Programs in Health Management Huntington, West Virginia University of Wisconsin - Madison Advisory Panel members provide valuable guidance during the preparation of OTA reports. However, the presence of an individual on the Advisory Panel does not mean that individual agrees with or endorses the conclusions of this particular paper. 53 APPENDIX E: ACKNOWLEDGMENTS

The author would like to express her appreciation to the following people for providing information and assistance.

Paul Anderson Diana DeAre Bureau of Emergency Medical Services Population Division Boise, Idaho U.S. Bureau of the Census Washington, DC Susan Baker The Johns Hopkins University Denise Denton School of Hygiene and Public Health Rural Health Office Baltimore, Maryland University of Arizona Tucson, Arizona Keith Bea Congressional Research Service Terry Dimon Washington, DC Memorial Hospital of Laramie County Cheyenne, Wyoming Calvin Beale Economic Research Service Gar Elison U.S. Department of Agriculture Community Health Services Washington, DC Utah Department of Health Salt Lake City, Utah Viola Behn New York Department of Health Bob Erikson Albany, New York Department of Geography University of Maryland, Baltimore County Richard Bohrer Baltimore, Maryland Division of Primary Care Services Health Resources and Services Administration Richard L. Forstall Rockville, Maryland Population Division U.S. Bureau of the Census James T. Bonnen Washington, DC Department of Agricultural Economics Michigan State University George Garnett East Lansing, Michigan National Association of EMS Physicians Soldotna, Alaska Don Dahmann Population Division Wilbert Gesler U.S. Bureau of the Census Department of Geography Washington, DC University of North Carolina Chapel Hill, North Carolina Robert Dawson National Rural Health Network Maria Gonzales National Rural Electric Cooperative Association Office of Management and Budget Washington, DC Washington, DC

54 Defining “Rural” Areas: Impact on Health Care Policy and Research • 55

Mary Goodwin Steven Rosenberg Department of Agriculture Salinas, California Austin, Texas Howard V. Stambler Rena Gordon Bureau of Health Professions Phoenix Rural Health Office Health Resources and Services Administration Phoenix, Arizona Rockville, Maryland

John Johnson Jack L. Stout Porter Memorial Hospital The 4th Party Inc. Valparaiso, Indiana Miami, Florida

Harvey Licht Elsie M. Sullivan New Mexico Health and Division of Primary Care Services Environment Department Health Resources and Services Administration Santa Fe, New Mexico Rockville, Maryland

David McGranahan Tom Swan Economic Research Service Rural E.M.S. Development Service U.S. Department of Agriculture Crested Butte, Colorado Washington, DC Robert T. Van Hook Carol Miller National Rural Health Association Mountain Management Kansas City, Missouri Ojo Sarco, New Mexico John Wardwell Steven Phillips Washington State University National Governor’s Association Pullman, Washington Washington, DC Barbara Wynn Jerome Pickard Bureau of Eligibility, Reimbursement, Appalachian Regional Commission and Coverage Washington, DC Health Care Financing Administration Baltimore, Maryland Stuart A. Reynolds Northern Montana Surgical Associates Havre, Montana

Thomas C. Ricketts Health Services Research Center University of North Carolina Chapel Hill, North Carolina REFERENCES

1. Abt Associates Inc., Assimin~ HosDitals to U rbanized Areas Within MetroDolitan Statisti- gal Areu, prepared for the Prospective Payment Assessment Commission, Washington, DC, June 1988.

2. Baker, S. P., Whitfield, R. A., and O’Neill, B., “County Mapping of Injured Mortality,” The Journal of Trauma 28(6):741-745, June 1988.

3. Baker, S. P., Whitfield, R. A., and O’Neill, B., “Geographic Variations in Mortality From Motor Vehicle Crashes,” New Endand Jou rnal of Medicine 316(22):1384-1387, May 28, 1987.

4. Bea, K., Congressional Research Service, Library of Congress, U.S. Congress, “Metropolitan Statistical Area Standards, History and Applications,” draft, Washington, DC, 1989.

5. Beale, C. L., Senior Demographer, Economic Research Service, U.S. Department of Agri- culture, Washington, DC, personal communication, May 1989.

6. Beale, C. L., “Poughkeepsie’s Complaint or Defining Metropolitan Areas,” American Demomar)hic~, pp. 29-48, January 1984.

7. Bender, L. D., Green, B. L., Hady, T. F., et al., Economic Research Service, U.S. Depart- ment of Agriculture, The Diverse Social and Economic Structure of Nonmetro r301itan America, Rural Development Research Report No. 49 (Washington, DC: U.S. Government Printing Office, September 1985).

8. Bonnen, J. T., Department of Agricultural Economics, Michigan State University, “The Statistical Data Base for Rural America,w No. 88-80, East Lansing, MI, October 1988.

9. Dahmann, D., Population Division, Bureau of the Census, U.S. Department of Commerce, Washington, DC, personal communication, May 1989.

10. De la Torre, A., Luft, H., University of California at San Francisco, and Fickenscher, K., Center for Medical Studies, Michigan State University, Kalamazoo, MI, “Zips Make a Difference: Methods to Improve Identification of Rural Subgroups,” draft for Pew Writing Seminar, Nov. 30, 1987.

110 Elison, G., “Frontier Areas: Problems for Delivery of Health Care Services,” Rural Health ,C~ 8(5):1-3, September/October 1986.

120 Forstall, R., Population Division, Bureau of the Census, U.S. Department of Commerce, Washington, DC, personal communication, May 19, 1989.

13. Gilford, D. M., Nelson, G. L., and Ingram, L. (eds.), Rural America in Passage: Statistic$ for Policy (Washington, DC: National Academy Press, 1981).

14. Hersh, A. S., and Van Hook, R. T., “A Research Agenda for Rural Health Services,” Health Services Research 23(6):1053- 1064, February 1989.

56 Defining “Rural” Areas: Impact on Health Care Policy and Research Ž 57

15. Hillen, S., “Medicare Underpays, Rural Missouri Hospitals Say,” American Medical News 31(10):27-28, Mar. 11, 1988.

16. Jenks, S., “Bill Would End Urban-Rural Hospital Pay Differential,” Medical World News 30(5):64, Mar. 13, 1989.

17. Joseph, A. E., and Phillips, D. R., “Measuring the Potential Physical Accessibility of Gener- al Practitioner Services,” Accessibility and Utilization: Geo~rar)hical perst)ect ives on Health Care Delivery (London: Harper and Row Ltd., 1984).

18. Kindig, D. A., Movassaghi, H., Chun, L. M., et al., University of Wisconsin, “Non- metropolitan County Typology and Health Resources,” unpublished manuscript, Madison, WI, WC. 15, 1988.

19. Licht, H., Program Manager, Primary Care Section, Public Health Division, New Mexico State Health and Environment Department, Santa Fe, NM, personal communication, May 1989.

20. Long, L., and DeAre, D., “U.S. Population Redistribution: A Perspective on the Non- metropolitan Turnaround,” Pom.dation and Develor)ment Review 14(3):433-450, September 1988.

21. Long, L., and DeAre, D., “Repopulating the Countryside: A 1980 Census Trend,” Science 217:1 111-1116, Sept. 17, 1982.

22. McGranahan, D. A., Hession, J. C., Hines, F. K., et al., Economic Research Service, U.S. Department of Agriculture, Soc ial and Economic Charac teristics of the PoDulation in . Metro and Nonmetro Cou t es. 1970-80 Rural Development Research Report No. 58 (Washington, DC: U.S. Go~~rnment Prin’ting Office, September 1986).

23. McKenzie, B. Y., National Mapping Division, U.S. Geological Survey, and LaMacchia, R. A., Geography Division, Bureau of the Census, U.S. Department of Commerce, “The U.S. Geological Survey -U.S. Bureau of the Census Cooperative Digital Mapping Project: A Unique Success Story,” Washington, DC, October 1987.

24. Meade, M. S., “Health Care Resources,” Medical GeoEraDhv (New York, NY: The Guilford Press, 1988).

25. Movassaghi, H., University of Wisconsin, Madison, WI, personal communication, May 1989.

26. New Mexico State Health and Environment Department, Public Health Division, Primary Care Section, “Frontier Areas in New Mexico-- 1987,” unpublished concept paper, Santa Fe, NM, 1987.

27. Patton, L., “Setting the Rural Health Services Research Agenda: The Congressional Per- spective,” Health Services Research 23(6): 1005-1052, February 1989.

28. Pickard, J,, “A New County Classification System,” Arlf)alachi~ 21(3):19-24, Summer 1988. 58 ■ Defining “Rural” Areas: Impact on Health Care Policy and Research

29. Pickard, J., “An Economic Development County Classification for the United States and Its Appalachian County Types,” Appalachian Regional Commission, Washington, DC, June 1988.

30. Popper, F. J., “The Strange Case of the Contemporary American Frontier,” Yale Review, pp. 101-121, Autumn 1986.

31. Prospective Payment Assessment Commission, Rer)ort and Reco mmendat ions to the Sec e- rtment of Health and Human Service s (Washington, DC: U.S. Governme~t nting Office, Apr. 1, 1987).

32. Prospective Payment Assessment Commission, Technical ADD endixes to the Rer)ort am$ Recommendations to the Sec eta Y. U.S. Det)a t e tof Health and Human Service$ (Washington, DC: U.S. Gove~nm~nt Printing ~f~c~, Apr. 1, 1987).

33. Prospective Payment Assessment Commission, ReDort and Reco me dati ons to the Sec Q- @ v. U.S. Det)artment of Health and Human Services (Washingt~n, & U.S. Governme~t P~nting Office, Mar. 1, 1988).

34. Prospective Payment Assessment Commission, “An Evaluation of the Department of Health and Human Services’ Report to Congress on Studies of Urban-Rural and Related Geographical Adjustments in the Medicare Prospective Payment System,” Washington, DC, June 1988.

35. Ricketts, T. C., Director, Program in Health Policy Analysis, Health Services Research Center, University of North Carolina, Chapel Hill, NC, personal communication, May 1989.

36. Rosenberg, S., “Rural Health Clinics Act Certification Benefits,” Rural Health Care 10(3):7 May/June 1988

37. Ross, P. J., Economic Research Service, U.S. Department of Agriculture, Washington, DC, “Remarks on the Development of a Policy-Oriented Classification of Nonmetropolitan Counties,” from proceedings of “Rural People and Places: A Symposium on Typology,” Grantville, PA, Oct. 22-24, 1986.

38. Schlosberg, J., “The MSA Mess,” American Demo~raDhi~, pp. 53-58, January 1989.

39. Sinclair, B., and Manderscheid, L. V., Department of Agriculture Economics, Michigan State University “A Comparative Evaluation of Indexes of Rurality--Their Policy Implica- tions and Distributional Impacts,” Research and Development Grant No. 21-26-73-52, prepared for the Manpower Administration, U.S. Department of Labor, Washington, DC, 1974.

40. Smith, B. J., and Parvin, D. W., Jr., “Defining and Measuring Rurality,” so uthern Jou rnal of AEricu Itural Economics, pp. 109-113, July 1973.

41. Speaker, R., Population Division, Bureau of the Census, U.S. Department of Commerce, Washington, DC, personal communication, May 1989. Defining “Rural” Areas: Impact on Health Care Policy and Research • 59

42. Stambler, H., Director, Office of Data and Management, Bureau of Health Professions, Health Resources and Services Administration, Public Health Service, U.S. Department of Health and Human Services, Rockville, MD, personal communication, March 1989.

42a. Thomas, G. S., “Micropolitan America,” American Demo~raDhic~, 11(5):20-24, May 1989.

43. U.S. Congress, House of Representatives, Task Force on the Rural Elderly of the Select Committee on Aging, Statu s of the Rural Elderly. Vol. I, Comm. Pub. No. 98-397 (Wash- ington, DC: U.S. Government Printing Office, 1983).

44. U.S. Congress, House of Representatives, Committee on Post Office and Civil Service, DesijmatinE Morgan and Lawrence Cou nties in Alabama as a Sinde MetroDo litan Statisti- cal Area, H. Rpt. 100-503 (Washington, DC: U.S. Government Printing Office, Feb. 9, 1988).

45. U.S. Congress, House of Representatives, “Designating Morgan and Lawrence Counties in Alabama as a Single Metropolitan Statistical Area,” Con~ressional Record, vol. 134, Mar. 1 and 2, 1988.

46. U.S. Congress, Senate, Familv Health Service s Amendments Act of 1988, S. Rpt. 100-343 (Washington, DC: U.S. Government Printing Office, May 13, 1988).

47. U.S. Department of Commerce, Bureau of the Census, Historical Statistics of the U. S., Colonial Times to 1970. Bicentennial Edition. Part 2 (Washington, DC: U.S. Government Printing Office, 1975).

48. U.S. Department of Commerce, Bureau of the Census, “Geography for a Changing Society,” Ce nsus 80: continuin~ Factfinder Tradition (Washington, DC: U.S. Government Printing Office, September 1980).

49. U.S. Department of Commerce, Bureau of the Census, census of PODU lation. 1980; Volume 1. Characteristics of the PODU lation: General Social and Economic Characte ristic~ (Washington, DC: U.S. Government Printing Office: 1981).

50. U.S. Department of Commerce, Bureau of the Census, Census o f Population and Housin~ 1980: Public-Use Microdata Sam~les. Technical Documentation, (Washington, DC: U.S. Government Printing Office, March 1983).

51. U.S. Department of Commerce, Bureau of the Census, County and City Data Book: 1983 (Washington, DC: U.S. Government Printing Office, 1983).

52. U.S. Department of Commerce, Bureau of the Census, “Census and Geography-Concepts and Products,” Factfinder CFF No. 8 (Washington, DC: U.S. Government Printing Office, August 1985).

53. U.S. Department of Commerce, Bureau of the Census, ‘Data for Small Communities,” Factfinder CFF No. 22 (Rev.) (Washington, DC: U.S. Government Printing Office, Janu- ary 1986).

54. U.S. Department of Commerce, Bureau of the Census, statistical Abstract of the U nited States: 1988, 108th Edition (Washington, DC: U.S. Government Printing Office, 1987). 60 • Defining ‘Rural” Areas: Impact on Health Care Policy and Research

55. U.S. Department of Commerce, Bureau of the Census, and U.S. Department of Agricul- ture, Economic Research Service, “Rural and Rural Farm Population: 1987,” ~urrent Pop- ulation ReDorts: Farm PODU lation, Series P-27, No. 61 (Washington, DC: U.S. Government Printing Office, June 1988).

56. U.S. Department of Commerce, Office of Federal Statistical Policy and Standards, Federal Committee on Standard Metropolitan Statistical Areas, “The Metropolitan Statistical Area Classification: 1980 Official Standards and Related Documents,w reprints from Statistical ReDo rter, December 1979 and August 1980.

57. U.S. Department of Commerce, Social and Economic Statistics Administration, Bureau of the Census, Public Use SamDles of Basic Records From the 1970 Census: Des critXion and Technical Documentalion, issued April 1972, (Washington, DC: U.S. Government Printing Office, 1974).

58. U.S. Department of Health and Human Services, ReDort to Con~ress: Stud ies of Urban- Rural and Related Geo~raDhical Ad iust ments in the Medicare ProsDect ive Payment !Wq- W, Washington, DC, Dec. 24, 1987.

59. U.S. Department of Health and Human Services, Public Health Service, Health Resources and Services Administration, Bureau of Health Care Delivery and Assistance, “Primary Care Activities in Frontier Areas - Regional Program Guidance Memorandum 86- 10,” unpublished memorandum, Rockville, MD, June 10, 1986.

60. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health, Atlas of U.S Caricer Mortalitv Amomz Whites: 1950- 198Q, DHHS Pub. No. (NIH) 87-2900 (Bethesda, MD: 1987).

61. U.S. Department of Health and Human Services, Public Health Service, Health Resources and Services Administration, Bureau of Health Professions, Office of Data Analysis, ~ Area Resource File (ARF) Svstem Information for Health Resources Planning and Re- $ea rch, ODAM Report No. 6-88 (Springfield, VA: National Technical Information Service, June 1988).

62. U.S. Executive Office of the President, Office of Management and Budget, bulletins revising definitions and designations of MSAS, nos. 83-20, 84-16, 84-24, 85-18, 86-14, 86-14, and 88-14, Washington; DC, June 27, 1983, June 29, 1984, Oct. 23, 1984, June 27, 1985, June 13, 1986, Nov. 19, 1986, and June 23, 1988.

63. Van Hook, R., Executive Director, National Rural Health Association, Kansas City, KS, personal communication, May 1989.

64. Wright, G. E., and Baily, S., SysteMetrics/McGraw-Hill, “Urban and Rural Cost Dif- ferences: Literature Synthesis and Review,” Technical Report #E-89-01, prepared for the Prospective Payment Assessment Commission, Washington, DC, February 1989.

65. Wynn, B., Director, Division of Hospital Payment Policy, Bureau of Eligibility, Reimbur- sement, and Coverage, Health Care Financing Administration, Washington, DC, personal communication, May 1989. 0